Anthropic, released Feb 17, 2026

Claude Sonnet 4.6price, context, benchmarks and release details

Provisional: not enough results to rank yet 100% confidence 100 percent, Full confidence, 5 of 7 expected sources in
68.3
SI Score
#20 of 142 ranked models
Input, per 1M tokens
$3.00Anthropic API pricingOfficial Claude API base tokens; lowest short-context global Standard rate; cache, batch, fast and regional premiums excluded Retrieved Oct 9, 2026 · factual citation
Open source ↗
Output, per 1M tokens
$15.00Anthropic API pricingOfficial Claude API base tokens; lowest short-context global Standard rate; cache, batch, fast and regional premiums excluded Retrieved Oct 9, 2026 · factual citation
Open source ↗
Context window
1Mmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Max output
64Kmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Released
Feb 17, 2026models.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗

How the score breaks down

Coding (weight 40 percent) 60.6
Math (weight 15 percent) 81.9
Preference (weight 15 percent) 78.4
Reasoning (weight 30 percent) 80.9

Weights: reasoning 30%, math 15%, coding 40%, preference 15%. Results use fixed 0–100 scales before averaging, and thin evidence is pulled toward 50. Method si-v3-retained-evidence-2, computed Oct 9, 2026, 05:13 UTC.

Around it on the leaderboard

  1. 18 Qwen3.8 Max 68.7
  2. 19 Gemini 3.5 Flash 68.3
  3. 20 Claude Sonnet 4.6 68.3
  4. 21 GPT-5.6 Terra 68.2
  5. 22 Gemini 3.1 Pro Preview 68.1

Full leaderboard

Benchmark results

27 benchmarks, 37 results

Each row shows the best published result. Where a model was tested at several settings, such as reasoning effort, open the row to see each one. Hover or tap a value for its source.

ARC-AGI-1 (public eval)reasoning 95.8%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'max' output effort. [variant] claude_sonnet_4_6_maxPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 95.8 2 settings
  • Setting 1 95.3%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
  • Setting 2 95.8%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'max' output effort. [variant] claude_sonnet_4_6_maxPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
About ARC-AGI-1 (public eval)
ARC-AGI-1 (semi-private)reasoning 86.5%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 86.5 2 settings
  • Setting 1 86.5%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
  • Setting 2 86%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'max' output effort. [variant] claude_sonnet_4_6_maxPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
About ARC-AGI-1 (semi-private)
ARC-AGI-2 (public eval)reasoning 65.7%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 65.7 2 settings
  • Setting 1 65.7%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
  • Setting 2 62.4%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'max' output effort. [variant] claude_sonnet_4_6_maxPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
About ARC-AGI-2 (public eval)
ARC-AGI-2 (semi-private)reasoning 60.4%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 60.4 2 settings
  • Setting 1 60.4%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'high' output effort. [variant] claude_sonnet_4_6_highPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
  • Setting 2 58.3%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. Evaluation conducted with a 120K thinking budget and a maximum context window of 128K tokens and 'max' output effort. [variant] claude_sonnet_4_6_maxPublished Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
    Open source ↗
About ARC-AGI-2 (semi-private)
GPQA Diamondreasoning 87.4%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] 32KPublished Feb 20, 2026 Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 87.4 4 settings
  • medium effort 83.3%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] mediumPublished Jul 13, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
  • high effort 83.3%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] highPublished Jul 13, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
  • max effort 78.8%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] maxPublished Aug 6, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
  • 32K thinking 87.4%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] 32KPublished Feb 20, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
About GPQA Diamond
Humanity's Last Examreasoning 34.6%Official model cards via models.devLab-reported; metric accuracy; transcribed by MIT models.dev catalog; not independently evaluated [variant] no toolsPublished Jun 30, 2026 Retrieved Oct 9, 2026 · factual citation; MIT transcription
Open source ↗ 34.6
Humanity's Last Exam (with tools)reasoning 46.8%Official model cards via models.devLab-reported; metric accuracy; transcribed by MIT models.dev catalog; not independently evaluated [variant] with toolsPublished Jun 30, 2026 Retrieved Oct 9, 2026 · factual citation; MIT transcription
Open source ↗ 46.8
LiveBench Reasoning: connectionsreasoning 99.3%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 99.3
LiveBench Reasoning: consecutive eventsreasoning 92.7%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 92.7
LiveBench Reasoning: logic with navigationreasoning 70%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 70.0
LiveBench Reasoning: spatialreasoning 100%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 100.0
LiveBench Reasoning: theory of mindreasoning 73.1%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 73.1
LiveBench Reasoning: zebra puzzlesreasoning 96%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 96.0
LiveBench Math: AMPS Hardmath 76%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 76.0
LiveBench Math: competition mathmath 94.1%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 94.1
LiveBench Math: integralsmath 90%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 90.0
LiveBench Math: olympiadmath 87.9%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 87.9
LiveBench Math: simplifymath 62.9%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 62.9
OTIS Mock AIME 2024–2025math 85.8%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] 32KPublished Feb 20, 2026 Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 85.8 4 settings
  • medium effort 82.2%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] mediumPublished Jul 13, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
  • high effort 75.6%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] highPublished Jul 13, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
  • max effort 71.1%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] maxPublished Aug 6, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
  • 32K thinking 85.8%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] 32KPublished Feb 20, 2026 Retrieved Oct 9, 2026 · CC-BY
    Open source ↗
About OTIS Mock AIME 2024–2025
LiveBench Coding: code completioncoding 78.3%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 78.3
LiveBench Coding: code generationcoding 80.3%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 80.3
LiveBench Coding: JavaScriptcoding 54.5%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 54.5
LiveBench Coding: Pythoncoding 50%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 50.0
LiveBench Coding: TypeScriptcoding 23.3%LiveBenchLiveBench subtask result; dataset edition 2026_06_25; publication time is HTTP Last-Modified of the score CSV, not the dataset edition or individual evaluation date [variant] 2026_06_25Published Oct 7, 2026 Retrieved Oct 9, 2026 · factual citation; Apache-2.0 code
Open source ↗ 23.3
SWE-bench Verifiedcoding 75.2%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Feb 21, 2026 Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 75.2
Terminal-Bench 2.1coding 67%Official model cards via models.devLab-reported; metric success rate; transcribed by MIT models.dev catalog; not independently evaluated [variant] Terminus-2; 2.1Published Jun 30, 2026 Retrieved Oct 9, 2026 · factual citation; MIT transcription
Open source ↗ 67.0
LMArena Textpreference 1457.7 eloLMArena / ArenaPublished source fact [variant] text / overallPublished Oct 8, 2026 Retrieved Oct 9, 2026 · CC-BY-4.0
Open source ↗ 78.4

Normalization uses a fixed 0–100 scale for each unit, independent of other models. Compare evaluation conditions before reading a small gap as decisive. “Lab-reported” marks the provider's own published figure.

Details and sources

Open weights
Nomodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
License
not yet reported
Input modalities
text, image, pdfmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
First seen by SuperIndex
Oct 8, 2026
Coverage
81% of expected source weight

Reported (5)

  • ARC PrizeOct 8, 2026
  • Epoch AI BenchmarkingOct 8, 2026
  • Official model cards via models.devOct 8, 2026
  • LiveBenchOct 8, 2026
  • LMArena / ArenaOct 8, 2026

Awaiting (2)

  • Humanity’s Last Exam13% of weight
  • Terminal-Bench6% of weight

Confidence rises as pending sources publish. Some sources never cover some models, so confidence reaches 100% at 80% of expected weight.

What changed

All releases · RSS feed

Alerts on this device

What to be alerted about
RSS feed