Corrections
Every published item correction and arithmetic note, including items older than the recent 100-item window. Edition corrections are counted separately.
41 item notices in this render · 1 edition correction records in the archive. Item counts and the item median below do not include editions.
41 item corrections or arithmetic notes since 2026-09-22 00:02 UTC · median time to correct 85h 17m
A corrected item is left in place. The headline and the original
text are unchanged, with the correction stamped on them, and the item stays in
Corrections and minor arithmetic notes do not automatically define a review verdict. The accuracy ledger counts human review outcomes separately. Each item notice records a cause — what actually went wrong — and how long the claim stood before it was withdrawn. That second number is the one worth arguing about. Everybody corrects things eventually; the question is how fast.
data.json and in every feed with a corrected category, or arithmetic-note for a minor note. An item that can
quietly disappear makes every other item unfalsifiable, so none of them does.Corrections and minor arithmetic notes do not automatically define a review verdict. The accuracy ledger counts human review outcomes separately. Each item notice records a cause — what actually went wrong — and how long the claim stood before it was withdrawn. That second number is the one worth arguing about. Everybody corrects things eventually; the question is how fast.
CORRECTED Application Integration privilege escalation patched August 27
CORRECTED 2026-10-15 01:38 UTC — CORRECTION: Google's GCP-2026-068 bulletin states that CVE-2026-85298 in Application Integration was patched on August 27, 2026 and that no customer action is required. RADAR's ACTION NEEDED / VERIFY_EXPOSURE framing, the triage claim that platform leads must verify patch status, and the mandatory historical Cloud Audit Logs review in WHAT TO DO went beyond the source and should not be treated as Google requirements. Google requires no customer patch-status verification or historical log review for this bulletin. Optional RADAR defense-in-depth advice: teams with a specific concern about past service account misuse may choose to review relevant Cloud Audit Logs under their own security procedures. This optional advice is RADAR's, not Google's. The original article and source evidence are retained below as the historical record.
Google Cloud security bulletins · published 2026-10-14 17:13 UTC · corrected 2026-10-15 01:38 UTC · stood 8h 25m · cause writer overreach
WHAT IT SAIDGoogle patched CVE-2026-85298 in Application Integration on August 27, 2026. The vulnerability allowed authenticated users to escalate privileges and impersonate service accounts through incorrect authorization in AuthConfig validation and integration execution runtime.
CORRECTED DeepSeek V4 Flash completion price cut 25%, prompt price up 12.5%
CORRECTED 2026-10-14 02:35 UTC — CORRECTION: The thresholds of 53 completion tokens per prompt token in WHY IT MATTERS and 53:1 in WHAT TO DO are wrong. Using only the captured base rates for ~deepseek/deepseek-v4-flash-latest, break-even is exactly 1 completion token(s) per 120 prompt token(s); total request cost is lower above that completion-to-prompt ratio, equal at that ratio, and higher below it, excluding all other charges. This is one output token per 120 input tokens (output/input = 1/120, approximately 0.008333); equality is break-even, not savings. At ten completion tokens per prompt token, total request cost falls 24.9%, consistent with WHAT CHANGED. Use that corrected threshold when interpreting the original budget advice. Captured arithmetic only; provider rate validity, applicability, event date and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-10-13 00:42 UTC · corrected 2026-10-14 02:35 UTC · stood 25h 52m · cause arithmetic
WHAT IT SAIDOpenRouter's pricing for ~deepseek/deepseek-v4-flash-latest changed on 13 October 2026. The captured completion base rate changed from $0.768 to $0.576 per 1M tokens. The captured prompt base rate changed from $0.0128 to $0.0144 per 1M tokens. For workloads generating ten completion tokens for every prompt token, total cost per request decreased 24.9%.
CORRECTED DeepSeek V4 Flash 0731 prompt pricing dropped 55.7%
CORRECTED 2026-10-03 23:15 UTC — Correction — The original workload example overstated both dollar amounts by 1,000×. Using the rates quoted in this item, 100M prompt tokens per day cost $1.15 before and $0.51 after, excluding other charges. This verifies internal arithmetic only; historical source rate applicability and effective dates remain unverified.
OpenRouter model catalogue · published 2026-10-03 01:37 UTC · corrected 2026-10-03 23:15 UTC · stood 21h 38m · cause arithmetic
WHAT IT SAIDOpenRouter reduced prompt token pricing for deepseek/deepseek-v4-flash-0731 from $0.0115/M tokens to $0.0051/M tokens, effective now. Completion pricing remains $1.28/M tokens.
CORRECTED DeepSeek V4 Flash 0731 prompt pricing tripled to $0.0154/M tokens
CORRECTED 2026-10-03 23:15 UTC — Correction — The original workload example overstated both dollar amounts by 1,000×. Using the rates quoted in this item, 100M prompt tokens per day cost $0.51 before and $1.54 after, excluding other charges. This verifies internal arithmetic only; historical source rate applicability and effective dates remain unverified.
OpenRouter model catalogue · published 2026-10-03 05:20 UTC · corrected 2026-10-03 23:15 UTC · stood 17h 55m · cause arithmetic
WHAT IT SAIDOpenRouter increased the prompt token price for deepseek/deepseek-v4-flash-0731 from $0.0051/M to $0.0154/M tokens, a 202% increase. Completion pricing remains $1.28/M tokens.
CORRECTED DeepSeek V4 Flash prompt pricing increased 22% on OpenRouter
CORRECTED 2026-10-03 23:12 UTC — Correction — The original workload example overstated both dollar amounts by 1,000×. Using the rates quoted in this item, 500K prompt tokens cost $0.0077 before and $0.0094 after per request, excluding other charges. This verifies internal arithmetic only; historical source rate applicability and effective dates remain unverified.
OpenRouter model catalogue · published 2026-10-03 11:41 UTC · corrected 2026-10-03 23:12 UTC · stood 11h 31m · cause arithmetic
WHAT IT SAIDOpenRouter raised the prompt token price for deepseek/deepseek-v4-flash-latest from $0.0154/M to $0.0188/M tokens, a 22.1% increase. Completion pricing remains $1.28/M tokens.
CORRECTED Anthropic renamed tiers and moved Claude Code to Pro and Max
CORRECTED 2026-10-02 19:08 UTC — Correction: The pricing page was alternating between two layout variants (see 26e9939b), and this item read one swing as a change. No tiers were renamed: only the "Developer" tab was removed, and "Individual" remains a category header. Claude Code appearing under Pro and Max reflects one side of that alternation, not a change in which plans include it.
Anthropic pricing · published 2026-09-21 23:43 UTC · corrected 2026-10-02 19:08 UTC · stood 259h 25m · cause writer overreach
WHAT IT SAIDThe 'Individual' and 'Developer' tier labels were removed from the pricing page. Claude Code, previously listed only under Pro, now appears under both Pro ($17/month) and Max (from $100/month). The Max tier description changed from 'For people who work with Claude all day' to the same phrasing.
CORRECTED Anthropic renamed Individual tier to Team, moved Claude Code to Max only
CORRECTED 2026-10-02 19:07 UTC — Correction: The pricing page was alternating between two layout variants and this item read one swing as a change. No tier was renamed: only the "Developer" tab was removed, "Individual" is a category header, and "Team & Enterprise" already existed. Claude Code is included in both Pro and Max, not Max only. Four duplicates of this item were corrected on 22 Sep; this original was missed.
Anthropic pricing · published 2026-09-21 23:44 UTC · corrected 2026-10-02 19:07 UTC · stood 259h 23m · cause writer overreach
WHAT IT SAIDThe Individual pricing tier no longer exists; it has been renamed to Team. Claude Code, previously included in both Pro ($17/month) and Max (from $100/month), is now only available in the Max tier. The Pro tier description now lists Claude Code as a feature, but the Max tier no longer states 'Claude Code included' in its header.
CORRECTED DeepSeek V4 Flash 0731 prompt pricing drops 37% on OpenRouter
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “A system processing 100M prompt tokens monthly saves $630.” Fact: Using the quoted or captured rates, 100M prompt tokens per month saves $0.63/month, not $630. Vendor rate applicability remains unverified.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-10-01 12:37 UTC · corrected 2026-10-02 17:58 UTC · stood 29h 21m · cause arithmetic
WHAT IT SAIDOpenRouter reduced prompt token pricing for deepseek/deepseek-v4-flash-0731 from $0.0171/M tokens to $0.0108/M tokens, effective now. Completion pricing remains $1.28/M tokens.
CORRECTED DeepSeek V4.1 Flash prompt pricing reduced 21.5% on OpenRouter
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “A workload sending 100M prompt tokens per day saves $430/month.” Fact: Using the quoted rounded rates, 100M prompt tokens/day saves $0.43/day or $12.90 per 30 days. Using the saved raw diff after-rate $0.015543/M, it saves $0.4257/day or $12.771 per 30 days. Neither supports $430/month. Vendor rates and applicability remain unverified.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-10-01 04:38 UTC · corrected 2026-10-02 17:58 UTC · stood 37h 20m · cause arithmetic
WHAT IT SAIDOpenRouter reduced prompt token pricing for deepseek/deepseek-v4.1-flash from $0.0198/M tokens to $0.0155/M tokens, effective now. Completion pricing remains $0.396/M tokens.
CORRECTED DeepSeek V4 Flash pricing dropped 90% on completion tokens
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “Total cost per request drops 89-90% across all workload mixes. A workload generating ten completion tokens for every prompt token now costs $0.0014/M tokens instead of $0.0126/M tokens. Existing budget forecasts are now overstated by roughly 10x.” Fact: Completion pricing falls 89.5% and prompt pricing falls 17.5%; a universal 89–90% blended decrease is not supported. For 1M prompt tokens plus 10M completion tokens, captured cost is $12.5120 before and $1.3167 after (89.5% lower). Across all nonnegative token mixes, the decrease lies between 17.5% and 89.5%.
[CORRECTION · WHAT TO DO] Original: “you have 90% of it available for other work or can reduce the allocation.” Fact: Any available budget depends on actual token volumes and mix; recompute both components instead of assuming a 90% saving.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-30 22:00 UTC · corrected 2026-10-02 17:58 UTC · stood 43h 58m · cause arithmetic
WHAT IT SAIDOpenRouter reduced pricing for deepseek/deepseek-v4-flash-latest: completion tokens from $1.25/M to $0.13/M (-89.5%), prompt tokens from $0.012/M to $0.0099/M (-17.5%). Change is live as of 30 September 2026.
CORRECTED DeepSeek V4.1 Flash completion pricing up 146%, prompt down 43%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · headline] Original: “completion pricing up 146%” Fact: Completion pricing increases 36.6%.
[CORRECTION · WHAT CHANGED] Original: “completion tokens from $0.384/token to $0.944/token (+145.8%)” Fact: Completion pricing increases from $0.29/M to $0.396/M (+36.6%). The quoted values were derived from capacity fields, not actual prices.
[CORRECTION · WHY IT MATTERS] Original: “Any workload generating completions will cost 2.5x more per completion token. The prompt discount does not offset this: even at ten completion tokens per prompt token, total cost per request increases 145.8%. Budgets based on prior pricing are now materially wrong.” Fact: Completion price multiplies by 1.3655; prompt price decreases 43.4%. At ten completion tokens per prompt token, captured cost increases 35.6%, not 145.8%. Net impact depends on the mix.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-30 22:00 UTC · corrected 2026-10-02 17:58 UTC · stood 43h 58m · cause extraction error
WHAT IT SAIDOpenRouter's deepseek/deepseek-v4.1-flash model changed pricing: completion tokens from $0.384/token to $0.944/token (+145.8%), prompt tokens from $0.035/M to $0.0198/M (-43.4%). Maximum completion tokens increased from 384,000 to 943,718.
CORRECTED DeepSeek V4 Pro 0813 completion pricing dropped 58%, output limit cut in half
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · headline] Original: “completion pricing dropped 58%, output limit cut in half” Fact: Completion price decreases 43.3%; output capacity decreases 58.3%, more than half.
[CORRECTION · WHAT CHANGED] Original: “completion tokens from $943,718/M to $393,216/M (-58.3%)” Fact: Completion pricing falls from $3.49/M to $1.98/M (-43.3%). Those large integers are output-token limits.
[CORRECTION · WHY IT MATTERS] Original: “Workloads generating long outputs cost 58% less per request regardless of prompt size” Fact: Prompt pricing increases 67.4% while completion pricing decreases 43.3%; net cost depends on token mix. At ten completions per prompt, cost falls 42.0%.
[CORRECTION · triage.signals] Original: “pricing.prompt decreased 83% (0.0000003942 → 0.00000066)” Fact: Prompt pricing increases 67.4%, not decreases 83%.
[CORRECTION · triage.signals] Original: “pricing.prompt decreased 83% (0.0000003942 → 0.00000066)” Fact: The prompt-change magnitude is 67.4%.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-30 12:06 UTC · corrected 2026-10-02 17:58 UTC · stood 53h 52m · cause extraction error
WHAT IT SAIDOpenRouter's deepseek/deepseek-v4-pro-0813 model changed pricing: completion tokens from $943,718/M to $393,216/M (-58.3%), prompt tokens from $0.3942/M to $0.6600/M (+67.4%). Maximum completion tokens reduced from 943,718 to 393,216 tokens.
CORRECTED DeepSeek Pro Latest pricing increased 56% prompt, completion reduced 58%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · headline] Original: “completion reduced 58%” Fact: Completion pricing increases 78.7%.
[CORRECTION · WHAT CHANGED] Original: “completion tokens from $943,718/M to $393,216/M (-58.3%).” Fact: Completion pricing is $1.9584/M to $3.50/M (+78.7%). The numbers 943,718 and 393,216 are token limits.
[CORRECTION · WHY IT MATTERS] Original: “Cost per request drops 58% for all workloads regardless of prompt-to-completion ratio” Fact: Both captured prices rise: prompt +56.25%, completion +78.7173%; a fixed-volume blend rises between those values.
[CORRECTION · triage.signals] Original: “prompt pricing increased 15.6x” Fact: Prompt price becomes 1.5625 times its old value, a 56.25% increase.
[CORRECTION · triage.triage reasoning] Original: “Pricing more than doubled” Fact: Neither captured rate doubles: prompt rises 56.25% and completion 78.7173%.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-30 04:26 UTC · corrected 2026-10-02 17:58 UTC · stood 61h 32m · cause extraction error
WHAT IT SAIDOpenRouter changed pricing for deepseek/deepseek-pro-latest effective 2026-09-29: prompt tokens from $0.096/M to $0.150/M (+56.3%), completion tokens from $943,718/M to $393,216/M (-58.3%). Maximum completion tokens per request reduced from 943,718 to 393,216 tokens.
ARITHMETIC NOTE DeepSeek V4 Pro pricing increased 163% on OpenRouter
[ARITHMETIC NOTE · WHY IT MATTERS] Original: “goes from $24.36 to $64.14.” Fact: For 10M prompt and 30M completion tokens/month, captured cost is $24.36 before and $64.11552 after, approximately $64.12, not $64.14.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-30 04:25 UTC · corrected 2026-10-02 17:58 UTC · stood 61h 33m · cause arithmetic
WHAT IT SAIDOpenRouter raised pricing for deepseek/deepseek-v4-pro from $0.3480/M prompt tokens and $0.6960/M completion tokens to $0.9159/M and $1.8319/M respectively, effective now. The increase is 163.2% for both token types.
ARITHMETIC NOTE DeepSeek V3.2 Exp pricing doubled on OpenRouter
[ARITHMETIC NOTE · WHAT CHANGED] Original: “Total cost per request increases 102-103% across all workload mixes.” Fact: Across all nonnegative fixed token mixes, the price increase ranges from prompt-only 100.9% to completion-only 103.4%. At one, three and ten completions per prompt, it is approximately 102.4%, 102.9% and 103.2%.
[ARITHMETIC NOTE · WHY IT MATTERS] Original: “goes from $7.39 to $15.13.” Fact: At 10M prompt and 30M completion tokens/month, captured cost goes from $7.392 to $15.00.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-29 12:17 UTC · corrected 2026-10-02 17:58 UTC · stood 77h 41m · cause arithmetic
WHAT IT SAIDOpenRouter increased pricing for deepseek/deepseek-v3.2-exp effective now: prompt tokens from $0.1344/M to $0.2700/M (+100.9%), completion tokens from $0.2016/M to $0.4100/M (+103.4%). Total cost per request increases 102-103% across all workload mixes.
CORRECTED DeepSeek Flash Latest pricing and output limits changed
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHAT CHANGED] Original: “completion pricing rose from $384/M to $943.72/M (+145.8%)” Fact: Completion pricing rises from $0.39/M to $0.60/M (+53.8%). Those other values originate from output-token limits, not price fields.
[CORRECTION · WHY IT MATTERS] Original: “Total cost per request increases 145.8% across all workload mixes because completion tokens now cost more than twice as much, outweighing the prompt discount. Applications generating long outputs will see the largest absolute cost increase, though the higher token limit enables responses that were previously truncated.” Fact: Net cost depends on the token mix: prompt -66.7%, completion +53.8%. Do not infer a universal increase from token-limit growth. Higher limits are capacity, not dollar rates.
[CORRECTION · WHAT TO DO] Original: “$943.72/M completion” Fact: Use the captured completion rate of $0.60/M alongside $0.02/M prompt, subject to applicable source conditions.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-29 04:41 UTC · corrected 2026-10-02 17:58 UTC · stood 85h 17m · cause extraction error
WHAT IT SAIDOpenRouter updated deepseek/deepseek-flash-latest on 2026-09-29: prompt pricing dropped from $0.06/M to $0.02/M (-66.7%), completion pricing rose from $384/M to $943.72/M (+145.8%), and max_completion_tokens increased from 384,000 to 943,718 (+146%).
CORRECTED DeepSeek Flash Latest pricing increased 35-71% effective now
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · triage.signals] Original: “pricing.prompt decreased 71% (0.000000035 → 0.00000006)” Fact: pricing.prompt increased 71.4% (0.000000035 → 0.00000006).
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-28 23:02 UTC · corrected 2026-10-02 17:58 UTC · stood 90h 56m · cause arithmetic
WHAT IT SAIDOpenRouter raised pricing for deepseek/deepseek-flash-latest: prompt tokens from $0.035/M to $0.060/M (+71%), completion tokens from $0.290/M to $0.390/M (+35%). The change is live in the catalogue as of 28 September 2026.
CORRECTED DeepSeek V4 Flash Vision pricing more than doubled
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHAT CHANGED] Original: “Completion tokens increased 260% from $262,144/M to $943,717/M.” Fact: Completion pricing rises 104.1%, from $0.6468/M to $1.32/M. The numbers 262,144 and 943,717 are token limits.
[CORRECTION · WHY IT MATTERS] Original: “Total cost per request increases 260% regardless of prompt-to-completion ratio. A workload generating 100M completion tokens per month that cost $26,214 now costs $94,372. Budgets and cost projections based on the previous rate are now off by more than 2.5x.” Fact: Both captured price components rise 104.1%, so their fixed-usage blend rises 104.1% for every mix. At 100M completion tokens/month, the component goes from $64.68 to $132.00. Token-limit growth is not price growth.
[CORRECTION · triage.triage reasoning] Original: “output token limit tripled” Fact: Output limit increases from 262,144 to 943,717, approximately 3.60 times its previous value.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-28 04:08 UTC · corrected 2026-10-02 17:58 UTC · stood 109h 50m · cause extraction error
WHAT IT SAIDOpenRouter raised pricing for deepseek/deepseek-v4-flash-vision-exp effective 28 September 2026. Prompt tokens increased 104% from $0.2156/M to $0.4400/M. Completion tokens increased 260% from $262,144/M to $943,717/M. Maximum completion tokens per request increased from 262,144 to 943,717.
CORRECTED DeepSeek V3 prompt pricing dropped 19.6%, completion up 2.3%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · headline] Original: “completion up 2.3%” Fact: Completion price rises 15.6%, not 2.3%.
[CORRECTION · WHAT CHANGED] Original: “completion tokens from $16,384/M to $16,000/M (-2.3%).” Fact: Completion pricing is $0.89/M to $1.0287/M (+15.6%); 16,384 to 16,000 are output-token limits.
[CORRECTION · WHY IT MATTERS] Original: “Total cost per request decreases 2.3% across all workload mixes.” Fact: Net cost depends on the token mix: prompt pricing falls 19.6% while completion pricing rises 15.6%.
[CORRECTION · WHAT TO DO] Original: “Reforecast monthly spend downward by approximately 2.3% for this model.” Fact: Recalculate prompt and completion costs separately at the captured rates; do not assume a universal decrease.
[CORRECTION · triage.signals] Original: “prompt pricing increased ~19.5%” Fact: Prompt pricing decreases 19.6%.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-27 16:33 UTC · corrected 2026-10-02 17:58 UTC · stood 121h 25m · cause extraction error
WHAT IT SAIDOpenRouter changed pricing for deepseek/deepseek-chat (DeepSeek V3) effective now: prompt tokens from $0.32/M to $0.2574/M (-19.6%), completion tokens from $16,384/M to $16,000/M (-2.3%). Maximum completion tokens reduced from 16,384 to 16,000.
CORRECTED DeepSeek V4 Flash completion token limit cut from 384k to 131k
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHAT CHANGED] Original: “Completion pricing dropped 65.9% ($384,000/M to $131,072/M tokens)” Fact: Completion pricing falls 0.2976%, from $0.09408/M to $0.0938/M. Output-token capacity falls 65.9%.
[CORRECTION · WHY IT MATTERS] Original: “The pricing drop is substantial” Fact: The captured price reduction is approximately 0.3%; the output-capacity reduction is 65.9%.
[ARITHMETIC NOTE · triage.signals] Original: “pricing.prompt decreased ~0.5%” Fact: Prompt pricing decreases 0.2976% (approximately 0.3%).
[ARITHMETIC NOTE · triage.signals] Original: “pricing.completion decreased ~0.5%” Fact: Completion pricing decreases 0.2976% (approximately 0.3%).
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-27 11:33 UTC · corrected 2026-10-02 17:58 UTC · stood 126h 24m · cause extraction error
WHAT IT SAIDOpenRouter's DeepSeek V4 Flash (deepseek/deepseek-v4-flash) reduced max_completion_tokens from 384,000 to 131,072 tokens. Completion pricing dropped 65.9% ($384,000/M to $131,072/M tokens) and prompt pricing dropped 0.3% ($0.047/M to $0.0469/M tokens).
CORRECTED DeepSeek Pro Latest pricing cut 79%, max completion tokens reduced 58%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHAT CHANGED] Original: “from $0.00000035 to $0.00000074844 per completion token” Fact: The captured completion rate decreases from $0.0000035 to $0.00000074844/token (78.6% lower); the published before rate was tenfold too low.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-26 10:57 UTC · corrected 2026-10-02 17:58 UTC · stood 151h 1m · cause arithmetic
WHAT IT SAIDOpenRouter's deepseek-pro-latest model pricing dropped from $0.00000035 to $0.00000074844 per completion token (79% decrease) and $0.00000025 to $0.00000024948 per prompt token (0.2% decrease), effective now. Maximum completion tokens reduced from 943,718 to 393,216 tokens.
CORRECTED DeepSeek V4.1 Flash pricing increased 3x prompt, 2x completion
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “A workload processing 1B tokens monthly goes from ~$660 to ~$1500.” Fact: For 1B total prompt-plus-completion tokens, captured component prices bound cost at $99–$600 before and $300–$1,200 after. Exact cost requires the token mix; $660 and $1,500 exceed the respective bounds.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-26 10:57 UTC · corrected 2026-10-02 17:58 UTC · stood 151h 1m · cause arithmetic
WHAT IT SAIDOpenRouter pricing for deepseek/deepseek-v4.1-flash changed from $0.000000099 to $0.0000003 per prompt token (3x increase) and $0.0000006 to $0.0000012 per completion token (2x increase). No effective date stated.
CORRECTED DeepSeek V4 Flash pricing reduced 7% prompt, 24% completion
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “dropped by roughly 20-25% depending on your prompt-to-completion ratio” Fact: At unchanged token volumes, the feasible saving is 6.6667% (prompt-only) to 23.8095% (completion-only). A 25% saving is impossible; use the actual prompt/completion mix.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-26 03:56 UTC · corrected 2026-10-02 17:58 UTC · stood 158h 2m · cause arithmetic
WHAT IT SAIDOpenRouter lowered pricing for deepseek/deepseek-v4-flash-latest from $0.0000000225 to $0.000000021 per prompt token (7% decrease) and from $0.00000042 to $0.00000032 per completion token (24% decrease), effective 26 September 2026.
CORRECTED DeepSeek Flash Latest: prompt pricing cut 53%, completion tokens capped at 384k
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · triage.triage reasoning] Original: “max output tokens cut by half” Fact: Maximum output tokens decrease 59.3%, from 943,718 to 384,000.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-26 03:56 UTC · corrected 2026-10-02 17:58 UTC · stood 158h 2m · cause arithmetic
WHAT IT SAIDOpenRouter's deepseek/deepseek-flash-latest model pricing dropped from $0.000000075 to $0.000000035 per prompt token (53% reduction) and $0.0000003 to $0.00000029 per completion token (3% reduction). The max_completion_tokens limit decreased from 943,718 to 384,000 tokens.
CORRECTED DeepSeek V4 Flash 0731 prompt pricing dropped 30%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “per-request cost just dropped by roughly a quarter to a third depending on your prompt-to-completion ratio” Fact: At unchanged token volumes, the feasible saving is 0% (completion-only) to 30% (prompt-only). A one-third saving is impossible at these rates. A 25% saving requires 160/3, approximately 53.33, prompt tokens per completion token.
[CORRECTION · WHY IT MATTERS] Original: “A workload doing 100M prompt tokens per day saves $900/day.” Fact: At the captured rates, 100M prompt tokens per day saves $0.90/day.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-26 03:56 UTC · corrected 2026-10-02 17:58 UTC · stood 158h 2m · cause arithmetic
WHAT IT SAIDOpenRouter reduced prompt token pricing for deepseek/deepseek-v4-flash-0731 from $0.00000003 to $0.000000021 per token, a 30% decrease. Completion pricing remains $0.00000032 per token.
CORRECTED DeepSeek Flash Latest pricing increased 87.5% prompt, 38.8% completion
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · headline] Original: “pricing increased 87.5% prompt, 38.8% completion” Fact: Prompt pricing rises 87.5%; completion pricing falls 38.8%.
[CORRECTION · triage.signals] Original: “pricing.prompt decreased: 0.00000004 → 0.000000075 (↓ 87.5%)” Fact: pricing.prompt increased 87.5% (0.00000004 → 0.000000075).
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-25 16:44 UTC · corrected 2026-10-02 17:58 UTC · stood 169h 14m · cause arithmetic
WHAT IT SAIDOpenRouter's deepseek/deepseek-flash-latest model pricing changed: prompt tokens from $0.00000004 to $0.000000075 per token (87.5% increase), completion tokens from $0.00000049 to $0.0000003 per token (38.8% decrease).
CORRECTED DeepSeek V4.1 Flash pricing increased 2-3x, output limit tripled
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHAT TO DO] Original: “multiply by 2.5-3x to forecast the new cost.” Fact: For unchanged token volumes, calculate both components: prompt price multiplies by 2.1429 and completion price by 2.8571. The blend is within those bounds, not universally 2.5–3 times.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-25 03:20 UTC · corrected 2026-10-02 17:58 UTC · stood 182h 38m · cause arithmetic
WHAT IT SAIDOpenRouter's deepseek/deepseek-v4.1-flash model pricing increased from $0.00000014 to $0.0000003 per prompt token (2.14x) and $0.00000042 to $0.0000012 per completion token (2.86x). Maximum completion tokens increased from 131,072 to 393,216.
CORRECTED DeepSeek Flash Latest pricing: prompt down 20%, completion up 300%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “A workload with 1:10 prompt-to-completion ratio now costs 2.6x more per request.” Fact: At one prompt token per ten completion tokens, captured total cost becomes 3.9373 times the old cost, a 293.7% increase.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-24 11:15 UTC · corrected 2026-10-02 17:58 UTC · stood 198h 42m · cause arithmetic
WHAT IT SAIDOpenRouter changed pricing for deepseek/deepseek-flash-latest effective now: prompt tokens decreased from $0.00000005 to $0.00000004 (20% reduction), completion tokens increased from $0.00000025 to $0.000001 (300% increase, 4x multiplier).
ARITHMETIC NOTE DeepSeek V4.1 Flash pricing up 50% prompt, 20% completion; output limit cut 58%
[ARITHMETIC NOTE · WHY IT MATTERS] Original: “Monthly costs will increase proportionally for all usage” Fact: The cost increase depends on token mix: prompt +50%, completion +20%. The stated 1B prompt plus 200M completion token workload correctly rises from $200 to $270/month (+35%).
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-23 21:12 UTC · corrected 2026-10-02 17:58 UTC · stood 212h 46m · cause arithmetic
WHAT IT SAIDOpenRouter's DeepSeek V4.1 Flash model pricing increased from $0.0000001 to $0.00000015 per prompt token (50% increase) and $0.0000005 to $0.0000006 per completion token (20% increase). Maximum completion tokens dropped from 943,718 to 393,216 tokens (58% reduction).
CORRECTED DeepSeek V4 Flash pricing increased 27% prompt, 45% completion
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · headline] Original: “pricing increased 27% prompt, 45% completion” Fact: Prompt pricing increases 26.7%; completion pricing decreases 45%.
[CORRECTION · triage.signals] Original: “pricing.prompt decreased ~21%” Fact: Prompt pricing increases 26.7%, not decreases 21%.
[CORRECTION · triage.signals] Original: “pricing.prompt decreased ~21%” Fact: The prompt change magnitude is 26.7%.
[CORRECTION · triage.signals] Original: “both token prices moved in same direction (down)” Fact: Prompt price rises while completion price falls.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-23 10:56 UTC · corrected 2026-10-02 17:58 UTC · stood 223h 2m · cause arithmetic
WHAT IT SAIDOpenRouter raised DeepSeek V4 Flash Latest pricing from $0.00000003 to $0.000000038 per prompt token (+27%) and from $0.000001 to $0.00000055 per completion token (-45%). Change is live as of 23 September 2026.
CORRECTED NVIDIA Nemotron 3.5 Lightning: prompt price +14%, output limit cut 44%
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHY IT MATTERS] Original: “($0.01 per 100K tokens)” Fact: The prompt-rate increase is $0.001 per 100K prompt tokens, or $0.01 per million.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-23 10:56 UTC · corrected 2026-10-02 17:58 UTC · stood 223h 2m · cause arithmetic
WHAT IT SAIDPrompt pricing increased from $0.00000007 to $0.00000008 per token (14% increase). Maximum completion tokens reduced from 235,929 to 131,072 tokens (44% reduction). Changes are live as of 23 September 2026.
ARITHMETIC NOTE DeepSeek Pro Latest: pricing cut, max completion tokens reduced 59%
[ARITHMETIC NOTE · triage.signals] Original: “prompt pricing decreased ~9.8%” Fact: Prompt pricing decreases 9.6731%, approximately 9.7%.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-22 16:28 UTC · corrected 2026-10-02 17:58 UTC · stood 241h 30m · cause arithmetic
WHAT IT SAIDPrompt pricing dropped from $0.000000624 to $0.00000056364 per token (~10% decrease). Completion pricing dropped from $0.00000288 to $0.00000169092 per token (~41% decrease). Maximum completion tokens reduced from 943,718 to 384,000 tokens (59% reduction). Changes are live as of 22 September 2026.
CORRECTED OpenAI gpt-oss-20b pricing cut 40% prompt, 31% completion; output capped at 32K tokens
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · WHAT CHANGED] Original: “to $0.00000018 per prompt token” Fact: The captured after prompt rate is $0.000000018/token ($0.018/M), a 40% decrease, not $0.00000018/token.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-22 16:27 UTC · corrected 2026-10-02 17:58 UTC · stood 241h 31m · cause arithmetic
WHAT IT SAIDOpenRouter's gpt-oss-20b model pricing dropped from $0.00000003 to $0.00000018 per prompt token (40% reduction) and $0.00000013 to $0.00000009 per completion token (31% reduction). Maximum completion tokens reduced from 117,964 to 32,768, effective now.
CORRECTED DeepSeek Pro Latest pricing increased 12% prompt, 72% completion
CORRECTED 2026-10-02 17:58 UTC — [CORRECTION · triage.triage reasoning] Original: “max output tokens nearly doubled” Fact: Maximum output tokens become 2.4576 times the prior value (+145.8%); they more than double.
[ARITHMETIC NOTE · triage.signals] Original: “pricing.completion increased ~71.6%” Fact: Completion pricing increases 71.8508%, approximately 71.9%.
[ARITHMETIC NOTE · WHAT CHANGED] Original: “(+71.6%)” Fact: Completion pricing increases 71.8508%, approximately 71.9%.
Captured arithmetic only; vendor rate validity, applicability and effective conditions remain unverified.
OpenRouter model catalogue · published 2026-09-22 11:06 UTC · corrected 2026-10-02 17:58 UTC · stood 246h 52m · cause arithmetic
WHAT IT SAIDOpenRouter's deepseek-pro-latest model pricing changed: prompt tokens from $0.000000558624 to $0.000000624 per token (+11.7%), completion tokens from $0.000001675872 to $0.00000288 per token (+71.6%). Max completion tokens increased from 384,000 to 943,718.
CORRECTED DeepSeek V4 Flash pricing changed: prompt down 25%, completion up 31%
CORRECTED 2026-09-27 04:35 UTC — Correction, 2026-09-27: the blended figure in "why it matters" was wrong. This item said a workload generating ten completion tokens for every prompt token "now costs approximately 24% more per request". From the item's own numbers it is 30.7% more: before, one prompt token plus ten completion tokens cost $0.00000323; after, $0.0000042225. The per-token figures in "what changed" are correct — prompt down 25%, completion up 31.25%. Only the blended number was wrong, and it was the one a reader would have taken to their finance team. RADAR now computes pricing arithmetic deterministically in workers/lib/pricing-math.ts and refuses to publish a pricing item containing any percentage it did not calculate; this exact text is blocked by that gate.
OpenRouter model catalogue · published 2026-09-25 21:12 UTC · corrected 2026-09-27 04:35 UTC · stood 31h 23m · cause arithmetic
WHAT IT SAIDOpenRouter adjusted pricing for deepseek/deepseek-v4-flash-latest effective 2026-09-25. Prompt tokens decreased from $0.00000003 to $0.0000000225 per token (25% reduction). Completion tokens increased from $0.00000032 to $0.00000042 per token (31% increase).
CORRECTED AWS removed CVE-2026-7191 QnABot security bulletin entirely
CORRECTED 2026-09-24 22:33 UTC — Correction: AWS did not withdraw this bulletin. The bulletin page is still live and lists the same CVEs. AWS's feed stopped including the bulletin text on 24 Sep, and RADAR misread that as a removal. The rule that caused this is fixed.
AWS security bulletins · published 2026-09-24 19:12 UTC · corrected 2026-09-24 22:33 UTC · stood 3h 21m · cause source changed format
WHAT IT SAIDAWS Security Bulletin 2026-020-AWS, published 27 April 2026, was completely removed on 24 September 2026. The bulletin documented CVE-2026-7191, an arbitrary code execution vulnerability in QnABot on AWS versions ≤7.2.4 affecting authenticated administrators. All technical details including exploitation method, affected components (Lambda, OpenSearch, S3, DynamoDB), and version scope were deleted.
CORRECTED AWS withdrew security bulletin 2026-014 covering three RCE vulnerabilities in RES
CORRECTED 2026-09-24 22:33 UTC — Correction: AWS did not withdraw this bulletin. The bulletin page is still live and lists the same CVEs. AWS's feed stopped including the bulletin text on 24 Sep, and RADAR misread that as a removal. The rule that caused this is fixed.
AWS security bulletins · published 2026-09-24 19:12 UTC · corrected 2026-09-24 22:33 UTC · stood 3h 21m · cause source changed format
WHAT IT SAIDAWS Security Bulletin 2026-014, published 6 April 2026, was removed from the feed on 24 September 2026. The bulletin documented CVE-2026-5707, CVE-2026-5708, and CVE-2026-5709 affecting Research and Engineering Studio (RES) versions through 2025.12.01. All three allowed remote code execution or privilege escalation: command injection via session names (CVE-5707), privilege escalation to instance profile permissions (CVE-5708), and command execution via FileBrowser API (CVE-5709). The entire advisory text, affected version ranges, and remediation guidance have been deleted.
CORRECTED Anthropic renamed Individual tier to Team, moved Claude Code to Max only
CORRECTED 2026-09-22 00:02 UTC — Duplicate of 26e9939b: the Anthropic pricing page was oscillating between two variants, producing a byte-identical diff every run. Suppressed by rule as of commit 917ce8a (oscillation / duplicate_diff). The underlying claim was later reviewed and found false: claude.com/pricing shows Claude Code in both Pro and Max, and "Individual" is a category header, not a tier. Publishing it five times compounded it.
Anthropic pricing · published 2026-09-21 23:45 UTC · corrected 2026-09-22 00:02 UTC · stood 18m · cause writer overreach
WHAT IT SAIDThe Individual pricing tier no longer exists; it has been renamed to Team. Claude Code, previously included in both Pro ($17/month) and Max (from $100/month), is now only included in the Max tier. The Pro tier description no longer lists Claude Code as a feature.
CORRECTED Anthropic renamed Individual tier to Team, moved Claude Code to Max only
CORRECTED 2026-09-22 00:02 UTC — Duplicate of 26e9939b: the Anthropic pricing page was oscillating between two variants, producing a byte-identical diff every run. Suppressed by rule as of commit 917ce8a (oscillation / duplicate_diff). The underlying claim was later reviewed and found false: claude.com/pricing shows Claude Code in both Pro and Max, and "Individual" is a category header, not a tier. Publishing it five times compounded it.
Anthropic pricing · published 2026-09-21 23:44 UTC · corrected 2026-09-22 00:02 UTC · stood 18m · cause writer overreach
WHAT IT SAIDThe Individual pricing tier no longer exists; it has been renamed to Team. Claude Code, previously included in both Pro ($17/month) and Max (from $100/month), is now only included in the Max tier. The Pro tier description no longer lists Claude Code as a feature.
CORRECTED Anthropic renamed Individual tier to Team, moved Claude Code to Max only
CORRECTED 2026-09-22 00:02 UTC — Duplicate of 26e9939b: the Anthropic pricing page was oscillating between two variants, producing a byte-identical diff every run. Suppressed by rule as of commit 917ce8a (oscillation / duplicate_diff). The underlying claim was later reviewed and found false: claude.com/pricing shows Claude Code in both Pro and Max, and "Individual" is a category header, not a tier. Publishing it five times compounded it.
Anthropic pricing · published 2026-09-21 23:44 UTC · corrected 2026-09-22 00:02 UTC · stood 18m · cause writer overreach
WHAT IT SAIDThe Individual pricing tier no longer exists; it has been renamed to Team. Claude Code, previously included in both Pro ($17/month) and Max (from $100/month), is now only included in the Max tier. The Pro tier description no longer lists Claude Code as a feature.
CORRECTED Anthropic renamed Individual tier to Team, moved Claude Code to Max only
CORRECTED 2026-09-22 00:02 UTC — Duplicate of 26e9939b: the Anthropic pricing page was oscillating between two variants, producing a byte-identical diff every run. Suppressed by rule as of commit 917ce8a (oscillation / duplicate_diff). The underlying claim was later reviewed and found false: claude.com/pricing shows Claude Code in both Pro and Max, and "Individual" is a category header, not a tier. Publishing it five times compounded it.
Anthropic pricing · published 2026-09-21 23:44 UTC · corrected 2026-09-22 00:02 UTC · stood 18m · cause writer overreach
WHAT IT SAIDThe Individual pricing tier no longer exists; it has been renamed to Team. Claude Code, previously included in both Pro ($17/month) and Max (from $100/month), is now only included in the Max tier. The Pro tier description no longer lists Claude Code as a feature.
Editions
A weekly edition is written once and never rewritten, so a correction to one is a record beside
it rather than an edit to it. These are counted separately from item corrections above: the
median-time-to-correct figure is about items, and an edition is a different kind of claim.
EDITION · COVERAGE · 2026-09-27 04:00 UTC
Correction, 2026-09-27: this edition said "Nothing material changed … that is the system working" for the week of 14–20 September. RADAR's interpretation stage was dead from 15 September 20:56 UTC — an expired model key — while the fetch stage kept running and reporting success. Changes detected after that were not read within the week, and 12 of them came back MATERIAL when interpretation was restored on 21 September. The run count, source count and cost below are correct. The claim that nothing material changed was not RADAR's to make.
- 18 changes were detected after the last time interpretation ran this week, so they were not read within it.
- Interpretation last worked 123 h before the week ended (2026-09-15 20:56 UTC). RADAR cannot say it watched a week it stopped reading partway through.
The edition itself is
unchanged — a correction sits beside the record, not inside it.