Benchmark Almanac

FrontierCode 2.0

Claude Opus 5.5 45.5% leads since Oct 2026 In use

Agents resolve 132 maintainer-built tasks from deliberately ambiguous prompts. Rubric checks ask whether a maintainer would merge the change.

Draws on 55% more repositories than 1.1 and has 501 blocking checks for thoroughness, against 168 in 1.1.

Made by Cognition Released Oct 2026 Fixing real repositories Tasks kept private Last in a launch post Oct 2026 Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Code A program or patch run against tests.
cognition.com
26
Models scored on its own leaderboard
1
Launch post lists it, from 1 lab

Claude Opus 5.5 leads with 45.5%

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

26 models. Scroll sideways for the rest.

  1. 45.5 Claude Opus 5.5
  2. 42.8 GPT-6 Astra
  3. 41.0 Claude Sonnet 5.5
  4. 40.0 Gemini 4 Carbon
  5. 39.0 GPT-6.1 Sol
  6. 34.3 GPT-6 Sol
  7. 32.4 Claude Fable 5.1
  8. 28.6 Grok 4.7
  9. 27.7 Claude Haiku 5.5
  10. 27.6 GPT-5.6 Sol
  11. 26.7 Claude Opus 5
  12. 26.3 Gemini 4 Argon
  13. 24.3 Muse Wonder
  14. 23.8 SWE-2
  15. 22.6 GPT-5.5
  16. 21.9 GPT-6 Luna
  17. 19.6 Mistral Large 4
  18. 19.3 DeepSeek V4.1 Flash 1017
  19. 17.8 GLM-5.3
  20. 17.7 Kimi K3
  21. 16.1 SWE-1.7
  22. 14.7 Claude Opus 4.8
  23. 14.2 GLM-5.3-Flash
  24. 3.21 Inkling-Small
  25. 2.90 Beam 1.0
  26. 1.34 SWE-1.6
  1. 1 Claude Opus 5.5 max 45.5%
  2. 2 GPT-6 Astra max 42.8%
  3. 3 Claude Sonnet 5.5 max 41%
  4. 4 Gemini 4 Carbon medium 40%
  5. 5 GPT-6.1 Sol max 39%
  6. 6 GPT-6 Sol max 34.3%
  7. 7 Claude Fable 5.1 xhigh 32.4%
  8. 8 Grok 4.7 xhigh 28.6%
  9. 9 Claude Haiku 5.5 max 27.7%
  10. 10 GPT-5.6 Sol max 27.6%
  11. 11 Claude Opus 5 xhigh 26.7%
  12. 12 Gemini 4 Argon high 26.3%
  13. 13 Muse Wonder high 24.3%
  14. 14 SWE-2 max 23.8%
  15. 15 GPT-5.5 xhigh 22.6%
  16. 16 GPT-6 Luna max 21.9%
  17. 17 Mistral Large 4 high 19.6%
  18. 18 DeepSeek V4.1 Flash 1017 max 19.3%
  19. 19 GLM-5.3 max 17.8%
  20. 20 Kimi K3 high 17.7%
  21. 21 SWE-1.7 na 16.1%
  22. 22 Claude Opus 4.8 max 14.7%
  23. 23 GLM-5.3-Flash max 14.2%
  24. 24 Inkling-Small max 3.21%
  25. 25 Beam 1.0 max 2.90%
  26. 26 SWE-1.6 na 1.34%
#ModelScoreSettingDateSource
1 Claude Opus 5.5 45.5% max, Claude Code Sep 2026 Official leaderboard
2 GPT-6 Astra 42.8% max, Codex Sep 2026 Official leaderboard
3 Claude Sonnet 5.5 41% max, Claude Code Sep 2026 Official leaderboard
4 Gemini 4 Carbon 40% medium, Devin CLI Oct 2026 Official leaderboard
5 GPT-6.1 Sol 39% max, Codex Sep 2026 Official leaderboard
6 GPT-6 Sol 34.3% max, Codex Sep 2026 Official leaderboard
7 Claude Fable 5.1 32.4% xhigh, Claude Code Sep 2026 Official leaderboard
8 Grok 4.7 28.6% xhigh, Grok Build Sep 2026 Official leaderboard
9 Claude Haiku 5.5 27.7% max, Claude Code Oct 2026 Official leaderboard
10 GPT-5.6 Sol 27.6% max, Codex Jul 2026 Official leaderboard
11 Claude Opus 5 26.7% xhigh, Claude Code Jul 2026 Official leaderboard
12 Gemini 4 Argon 26.3% high, Devin CLI Sep 2026 Official leaderboard
13 Muse Wonder 24.3% high, Muse Code Oct 2026 Official leaderboard
14 SWE-2 23.8% max, Devin Sep 2026 Official leaderboard
15 GPT-5.5 22.6% xhigh, Codex Apr 2026 Official leaderboard
16 GPT-6 Luna 21.9% max, Codex Sep 2026 Official leaderboard
17 Mistral Large 4 19.6% high, Mistral Vibe Oct 2026 Official leaderboard
18 DeepSeek V4.1 Flash 1017 19.3% max, Devin CLI Oct 2026 Official leaderboard
19 GLM-5.3 17.8% max, Devin CLI Aug 2026 Official leaderboard
20 Kimi K3 17.7% high, Devin CLI Jul 2026 Official leaderboard
21 SWE-1.7 16.1% na, Devin CLI Jul 2026 Official leaderboard
22 Claude Opus 4.8 14.7% max, Claude Code May 2026 Official leaderboard
23 GLM-5.3-Flash 14.2% max, Devin CLI Aug 2026 Official leaderboard
24 Inkling-Small 3.21% max, Devin CLI Jul 2026 Official leaderboard
25 Beam 1.0 2.90% max, Devin CLI Oct 2026 Official leaderboard
26 SWE-1.6 1.34% na, Devin CLI Apr 2026 Official leaderboard

Labs report scores for 1 model in their cards. We plot every source in the charts below.

The best score, 45.5%, costs $13 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6 Astra at xhigh comes within 4 points of the best for $4.54 a task, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

OpenAI Anthropic Google DeepMind Cost frontier One model at each effort level
0% 10% 20% 30% 40% 50% $0.01 $0.10 $1.00 $10 $100 Cost per task, log scale Claude Opus 5.5, max Claude Opus 5.5, xhigh Gemini 4 Carbon, medium GPT-6.1 Sol, high GPT-6.1 Sol, medium GPT-6.1 Sol, low Claude Haiku 5.5, xhigh 0% 10% 20% 30% 40% 50% $0.01 $0.10 $1.00 $10 $100 Cost per task, log scale Claude Opus 5.5, max GPT-6 Astra, xhigh
ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 max Claude Code 45.5% $13 201,919
Claude Opus 5.5 xhigh Claude Code 43.4% $5.18 88,053
GPT-6 Astra max Codex 42.8% $6.68 36,354
GPT-6 Astra xhigh Codex 41.4% $4.54 20,869
Claude Sonnet 5.5 max Claude Code 41% $20 514,722
Gemini 4 Carbon medium Devin CLI 40% $4.04 109,126
GPT-6.1 Sol max Codex 39% $1.03 41,479
GPT-6 Astra high Codex 38.3% $4.28 18,433
Gemini 4 Carbon high Devin CLI 37.5% $4.43 121,442
GPT-6 Astra medium Codex 36.8% $3.49 14,065
GPT-6.1 Sol high Codex 34.8% $0.65 22,419
Claude Sonnet 5.5 xhigh Claude Code 34.8% $3.31 80,433
Claude Opus 5.5 high Claude Code 34.5% $2.42 41,233
GPT-6 Sol max Codex 34.3% $2.35 54,181
Claude Fable 5.1 xhigh Claude Code 32.4% $8.68 92,813
GPT-6.1 Sol xhigh Codex 32.3% $0.69 24,789
Gemini 4 Carbon low Devin CLI 31.9% $4.55 57,868
Claude Fable 5.1 max Claude Code 31.7% $11 119,089
GPT-6 Astra low Codex 31.1% $2.44 8,987
GPT-6 Sol xhigh Codex 30.3% $1.44 28,587
Claude Opus 5.5 medium Claude Code 30.1% $1.70 30,068
GPT-6.1 Sol medium Codex 29.6% $0.49 15,686
Grok 4.7 xhigh Grok Build 28.6% $14 110,790
Claude Haiku 5.5 max Claude Code 27.7% $0.37 205,112
GPT-5.6 Sol max Codex 27.6% $5.50 38,352
Claude Opus 5 xhigh Claude Code 26.7% $11 76,436
Claude Opus 5 high Claude Code 26.7% $8.36 59,260
Claude Opus 5 max Claude Code 26.6% $13 88,196
Grok 4.7 high Grok Build 26.6% $11 94,341
Gemini 4 Argon high Devin CLI 26.3% $6.51 148,569
Grok 4.7 medium Grok Build 25.9% $8.03 72,180
Claude Fable 5.1 high Claude Code 25.7% $5.39 54,242
GPT-6.1 Sol low Codex 25.6% $0.34 10,042
GPT-5.6 Sol high Codex 24.3% $3.17 20,920
Muse Wonder high Muse Code 24.3% $3.14 32,832
Muse Wonder xhigh Muse Code 24.1% $3.79 38,672
GPT-5.6 Sol xhigh Codex 24.1% $3.83 26,415
Claude Opus 5 medium Claude Code 24% $5.22 39,813
SWE-2 max Devin 23.8% $1.26 97,584
Gemini 4 Argon medium Devin CLI 23.4% $5.69 123,879
Muse Wonder medium Muse Code 23.1% $2.53 26,714
SWE-2 high Devin 23.1% $0.96 68,546
GPT-6 Sol high Codex 22.8% $1.07 19,393
GPT-5.5 xhigh Codex 22.6% $5.28 30,925
Claude Haiku 5.5 xhigh Claude Code 22.3% $0.23 127,735
Claude Sonnet 5.5 high Claude Code 21.9% $1.07 27,839
GPT-6 Luna max Codex 21.9% $0.17 79,659
GPT-6 Sol medium Codex 21.6% $0.82 13,750
Claude Fable 5.1 medium Claude Code 21.5% $3.89 37,911
Gemini 4 Argon low Devin CLI 21.5% $3.98 63,291
Muse Wonder low Muse Code 21.2% $1.84 20,301
Grok 4.7 low Grok Build 20.8% $3.87 38,161
Claude Fable 5.1 low Claude Code 20.6% $3.01 28,797
Mistral Large 4 high Mistral Vibe 19.6% $1.55 88,069
GPT-5.6 Sol medium Codex 19.5% $2.36 15,446
DeepSeek V4.1 Flash 1017 max Devin CLI 19.3% $0.38 82,220
GPT-5.5 high Codex 19.1% $3.80 21,907
Claude Opus 5 low Claude Code 18.7% $3.28 26,898
Claude Haiku 5.5 high Claude Code 18.6% $0.14 80,993
GPT-6 Luna xhigh Codex 18.4% $0.13 50,565
DeepSeek V4.1 Flash 1017 high Devin CLI 18.2% $0.27 57,169
GLM-5.3 max Devin CLI 17.8% $7.52 100,706
Kimi K3 high Devin CLI 17.7% $4.47 49,854
Kimi K3 max Devin CLI 17.3% $5.46 64,271
GLM-5.3 high Devin CLI 16.8% $5.18 73,247
GPT-6 Luna high Codex 16.5% $0.12 42,170
SWE-1.7 na Devin CLI 16.1% $3.52 89,251
GPT-6 Luna medium Codex 16% $0.09 31,360
DeepSeek V4.1 Flash 1017 low Devin CLI 16% $0.21 40,659
SWE-2 medium Devin 15.6% $0.44 28,705
Claude Opus 4.8 max Claude Code 14.7% $12 125,863
Claude Opus 5.5 low Claude Code 14.7% $0.75 14,230
GLM-5.3 low Devin CLI 14.4% $3.19 47,199
Claude Opus 4.8 xhigh Claude Code 14.3% $9.13 88,364
GPT-5.5 medium Codex 14.2% $2.95 16,320
GLM-5.3-Flash max Devin CLI 14.2% $0.79 90,044
Claude Sonnet 5.5 medium Claude Code 14.2% $0.48 14,449
Claude Haiku 5.5 medium Claude Code 13.4% $0.10 57,582
GPT-5.6 Sol low Codex 12.8% $1.45 9,806
GLM-5.3-Flash high Devin CLI 12.4% $0.58 56,914
Kimi K3 low Devin CLI 11.1% $2.98 28,740
Claude Opus 4.8 medium Claude Code 11% $5.59 57,192
Claude Opus 4.8 high Claude Code 10.9% $6.32 62,298
GPT-6 Sol low Codex 10.5% $0.42 6,487
Claude Haiku 5.5 low Claude Code 10.5% $0.07 39,366
Claude Opus 4.8 low Claude Code 9.55% $4.20 40,687
Mistral Large 4 low Mistral Vibe 9.50% $0.67 28,384
Claude Sonnet 5.5 low Claude Code 9.41% $0.37 11,265
Mistral Large 4 medium Mistral Vibe 8.78% $1.03 51,238
GLM-5.3-Flash low Devin CLI 7.81% $0.43 35,321
GPT-5.5 low Codex 7.20% $1.98 10,332
GPT-6 Luna low Codex 4.05% $0.02 6,519
Inkling-Small max Devin CLI 3.21% $3.68 39,077
Beam 1.0 max Devin CLI 2.90% $1.30 27,971
Beam 1.0 xhigh Devin CLI 2.43% $1.13 23,156
Beam 1.0 high Devin CLI 1.74% $0.61 14,068
SWE-1.6 na Devin CLI 1.34% $1.36 23,990
Inkling-Small high Devin CLI 1.27% $1.37 13,658
Beam 1.0 medium Devin CLI 0.83% $0.32 7,243
Beam 1.0 minimal Devin CLI 0% $0.06 1,550
Beam 1.0 low Devin CLI 0% $0.14 3,490

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard 1 leader
Claude Opus 5.5 45.5%

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Claude Fable 5.1 GPT-5.6 Sol GPT-5.5 SWE-1.6 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Claude Fable 5.1

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Mistral AI
DateLaunch postScoreAs the post writes it
29 Oct 2026 Mistral Large 4 (Le Chonk, open weights) 19.7 Cognition FrontierCode 2.0

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

Anthropic xAI OpenAI
FrontierCode 2.0 7 leaders
Claude Opus 5.5 45.5%
FrontierCode 3 leaders
Claude Opus 5.5 54.6%
FrontierCode 1.1 3 leaders
GPT-6 Astra 64.5%
FrontierCode 2.0 1 leader
Claude Opus 5.5 45.5%
Replaces FrontierCode 1.1

More in Coding

All 243 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Opus 5.5 93.9%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471