Benchmark Almanac

Earth-Bench V2

Claude Opus 5.5 80% leads since October 2026
Not in model cards
Version 2

Agents analyze prepared satellite observations. The corrected release measures final answer accuracy, tool selection, argument accuracy, and execution efficiency.

The dataset supplies 496 entries across 248 underlying tasks. The authors publish new scores for GPT-5 and Qwen3-Max.

Released
October 2026
Kind
Apps and MCP servers
Level
University
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
Interaction
Agentic
Environment
APIs and tools
Items from
Real work and use
Run by
Anyone
Based on
Earth-Bench
In views
Agents, Vision
Strongest model tested
Claude Opus 5.5 #5 on the capability index
Last model added
October 2026
Scores refreshed
November 20th, 2026
48
Entries on Axis Spatial’s leaderboard, 47 models we track

Claude Opus 5.5 leads with 80% Claude Opus 5.5 leads with 80%

Best score per entry from Axis Spatial’s leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

  1. 1 Claude Opus 5.5 max 80%
  2. 2 GLM-5.3 max 77%
  3. 3 DeepSeek V4.1 Flash max 77%
  4. 4 Kimi K3 Reasoning 75%
  5. 5 GPT-6 Sol high 75%
  6. 6 Claude Opus 5 high 74%
  7. 7 DeepSeek-V4-Flash high 73%
  8. 8 Kimi K2.7 Code Reasoning 73%
  9. 9 Claude Fable 5.1 medium 73%
  10. 10 GPT-6.1 Sol medium 72%
  11. 11 Claude Haiku 5.5 high 72%
  12. 12 Claude Opus 4.8 high 71%
  13. 13 GLM-5.2 max 71%
  14. 14 Claude Sonnet 5.5 high 71%
  15. 15 GPT-5.6 Sol high 70%
  16. 16 Claude Opus 4.7 high 69%
  17. 17 DeepSeek-V4-Pro max 69%
  18. 18 Nemotron 3 Ultra high 69%
  19. 19 Qwen 3.8 27B Reasoning 69%
  20. 20 GLM-5.3-Flash max 69%
  21. 21 Kimi K2.6 Reasoning 68%
  22. 22 MiniMax-M3 Reasoning 68%
  23. 23 Claude Sonnet 5 high 68%
  24. 24 MiMo-V2.6 Flash Reasoning 68%
  25. 25 GPT-5.6 Luna high 67%
  26. 26 GPT-6 Luna xhigh 67%
  27. 27 Qwen 3.6 35B-A3B Reasoning 62%
  28. 28 GPT-5.6 Terra high 62%
  29. 29 Muse Glimmer high 60%
  30. 30 Step 5 Preview Reasoning 60%
  31. 31 Gemini 3.8 Flash low 54%
  32. 32 Gemini 3.1 Flash-Lite high 53%
  33. 33 Claude Haiku 4.5 Reasoning 51%
  34. 34 gpt-oss-120b high 50%
  35. 35 Mistral Large 4 high 50%
  36. 36 Nemotron 3.5 Super 120B Reasoning 49%
  37. 37 Gemma 4 26B A4B Reasoning 46%
  38. 38 Nemotron 3.5 Super VL Reasoning 46%
  39. 39 glm-4.7-flash Reasoning 44%
  40. 40 Laguna XS 2.1 Reasoning 44%
  41. 41 Gemma 4 31B IT Reasoning 42%
  42. 42 MiniMax-M3.1-Flash max 41%
  43. 43 gpt-oss-20b high 32%
  44. 44 Mistral Large 3 Non-reasoning 31%
  45. 45 Nemotron 3 Nano Omni 30B-A3B Reasoning 31%
  46. 46 Nemotron 3.5 Lightning 30B-A3B Reasoning 16%
  47. 47 Llama 3.3 70B Non-reasoning 9.00%
  48. 48 Nemotron 3.5 Bolt E4B Reasoning 5.00%

Source: Axis Spatial’s leaderboard

Source: Axis Spatial’s leaderboard

#EntryScoreSettingDateLinkSource
1 Claude Opus 5.5 80% max, Claude Code Sep 2026 axisspatial.com Official leaderboard
2 GLM-5.3 77% max, Claude Code Aug 2026 axisspatial.com Official leaderboard
3 DeepSeek V4.1 Flash 77% max, Claude Code Sep 2026 axisspatial.com Official leaderboard
4 Kimi K3 75% Reasoning, Terminus-2 Jul 2026 axisspatial.com Official leaderboard
5 GPT-6 Sol 75% high, Codex Sep 2026 axisspatial.com Official leaderboard
6 Claude Opus 5 74% high, Claude Code Jul 2026 axisspatial.com Official leaderboard
7 DeepSeek-V4-Flash 73% high, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
8 Kimi K2.7 Code 73% Reasoning, Kimi Code Jun 2026 axisspatial.com Official leaderboard
9 Claude Fable 5.1 73% medium, Claude Code Sep 2026 axisspatial.com Official leaderboard
10 GPT-6.1 Sol 72% medium, Codex Sep 2026 axisspatial.com Official leaderboard
11 Claude Haiku 5.5 72% high, Claude Code Oct 2026 axisspatial.com Official leaderboard
12 Claude Opus 4.8 71% high, Claude Code May 2026 axisspatial.com Official leaderboard
13 GLM-5.2 71% max, Claude Code Jun 2026 axisspatial.com Official leaderboard
14 Claude Sonnet 5.5 71% high, Claude Code Sep 2026 axisspatial.com Official leaderboard
15 GPT-5.6 Sol 70% high, Codex Jul 2026 axisspatial.com Official leaderboard
16 Claude Opus 4.7 69% high, Claude Code Apr 2026 axisspatial.com Official leaderboard
17 DeepSeek-V4-Pro 69% max, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
18 Nemotron 3 Ultra 69% high, Terminus-2 Jun 2026 axisspatial.com Official leaderboard
19 Qwen 3.8 27B 69% Reasoning, Qwen Code Aug 2026 axisspatial.com Official leaderboard
20 GLM-5.3-Flash 69% max, Claude Code Aug 2026 axisspatial.com Official leaderboard
21 Kimi K2.6 68% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
22 MiniMax-M3 68% Reasoning, MiniMax Code Jun 2026 axisspatial.com Official leaderboard
23 Claude Sonnet 5 68% high, Claude Code Jun 2026 axisspatial.com Official leaderboard
24 MiMo-V2.6 Flash 68% Reasoning, MiMo Code Sep 2026 axisspatial.com Official leaderboard
25 GPT-5.6 Luna 67% high, Codex Jul 2026 axisspatial.com Official leaderboard
26 GPT-6 Luna 67% xhigh, Codex Sep 2026 axisspatial.com Official leaderboard
27 Qwen 3.6 35B-A3B 62% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
28 GPT-5.6 Terra 62% high, Codex Jul 2026 axisspatial.com Official leaderboard
29 Muse Glimmer 60% high, Terminus-2 Aug 2026 axisspatial.com Official leaderboard
30 Step 5 Preview 60% Reasoning, Terminus-2 Sep 2026 axisspatial.com Official leaderboard
31 Gemini 3.8 Flash 54% low, Terminus-2 Sep 2026 axisspatial.com Official leaderboard
32 Gemini 3.1 Flash-Lite 53% high, Terminus-2 Mar 2026 axisspatial.com Official leaderboard
33 Claude Haiku 4.5 51% Reasoning, Claude Code Oct 2025 axisspatial.com Official leaderboard
34 gpt-oss-120b 50% high, Terminus-2 Aug 2025 axisspatial.com Official leaderboard
35 Mistral Large 4 50% high, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
36 Nemotron 3.5 Super 120B 49% Reasoning, Terminus-2 axisspatial.com Official leaderboard
37 Gemma 4 26B A4B 46% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
38 Nemotron 3.5 Super VL 46% Reasoning, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
39 glm-4.7-flash 44% Reasoning, Terminus-2 Jan 2026 axisspatial.com Official leaderboard
40 Laguna XS 2.1 44% Reasoning, Terminus-2 Jul 2026 axisspatial.com Official leaderboard
41 Gemma 4 31B IT 42% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
42 MiniMax-M3.1-Flash 41% max, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
43 gpt-oss-20b 32% high, Terminus-2 Aug 2025 axisspatial.com Official leaderboard
44 Mistral Large 3 31% Non-reasoning, Terminus-2 Dec 2025 axisspatial.com Official leaderboard
45 Nemotron 3 Nano Omni 30B-A3B 31% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
46 Nemotron 3.5 Lightning 30B-A3B 16% Reasoning, Terminus-2 Aug 2026 axisspatial.com Official leaderboard
47 Llama 3.3 70B 9.00% Non-reasoning, Terminus-2 Dec 2024 axisspatial.com Official leaderboard
48 Nemotron 3.5 Bolt E4B 5.00% Reasoning, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
#ModelScoreSettingDateLinkSource
1 Claude Opus 5.5 80% max, Claude Code Sep 2026 axisspatial.com Official leaderboard
2 GLM-5.3 77% max, Claude Code Aug 2026 axisspatial.com Official leaderboard
3 DeepSeek V4.1 Flash 77% max, Claude Code Sep 2026 axisspatial.com Official leaderboard
4 Kimi K3 75% Reasoning, Terminus-2 Jul 2026 axisspatial.com Official leaderboard
5 GPT-6 Sol 75% high, Codex Sep 2026 axisspatial.com Official leaderboard
6 Claude Opus 5 74% high, Claude Code Jul 2026 axisspatial.com Official leaderboard
7 DeepSeek-V4-Flash 73% high, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
8 Kimi K2.7 Code 73% Reasoning, Kimi Code Jun 2026 axisspatial.com Official leaderboard
9 Claude Fable 5.1 73% medium, Claude Code Sep 2026 axisspatial.com Official leaderboard
10 GPT-6.1 Sol 72% medium, Codex Sep 2026 axisspatial.com Official leaderboard
11 Claude Haiku 5.5 72% high, Claude Code Oct 2026 axisspatial.com Official leaderboard
12 Claude Opus 4.8 71% high, Claude Code May 2026 axisspatial.com Official leaderboard
13 GLM-5.2 71% max, Claude Code Jun 2026 axisspatial.com Official leaderboard
14 Claude Sonnet 5.5 71% high, Claude Code Sep 2026 axisspatial.com Official leaderboard
15 GPT-5.6 Sol 70% high, Codex Jul 2026 axisspatial.com Official leaderboard
16 Claude Opus 4.7 69% high, Claude Code Apr 2026 axisspatial.com Official leaderboard
17 DeepSeek-V4-Pro 69% max, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
18 Nemotron 3 Ultra 69% high, Terminus-2 Jun 2026 axisspatial.com Official leaderboard
19 Qwen 3.8 27B 69% Reasoning, Qwen Code Aug 2026 axisspatial.com Official leaderboard
20 GLM-5.3-Flash 69% max, Claude Code Aug 2026 axisspatial.com Official leaderboard
21 Kimi K2.6 68% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
22 MiniMax-M3 68% Reasoning, MiniMax Code Jun 2026 axisspatial.com Official leaderboard
23 Claude Sonnet 5 68% high, Claude Code Jun 2026 axisspatial.com Official leaderboard
24 MiMo-V2.6 Flash 68% Reasoning, MiMo Code Sep 2026 axisspatial.com Official leaderboard
25 GPT-5.6 Luna 67% high, Codex Jul 2026 axisspatial.com Official leaderboard
26 GPT-6 Luna 67% xhigh, Codex Sep 2026 axisspatial.com Official leaderboard
27 Qwen 3.6 35B-A3B 62% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
28 GPT-5.6 Terra 62% high, Codex Jul 2026 axisspatial.com Official leaderboard
29 Muse Glimmer 60% high, Terminus-2 Aug 2026 axisspatial.com Official leaderboard
30 Step 5 Preview 60% Reasoning, Terminus-2 Sep 2026 axisspatial.com Official leaderboard
31 Gemini 3.8 Flash 54% low, Terminus-2 Sep 2026 axisspatial.com Official leaderboard
32 Gemini 3.1 Flash-Lite 53% high, Terminus-2 Mar 2026 axisspatial.com Official leaderboard
33 Claude Haiku 4.5 51% Reasoning, Claude Code Oct 2025 axisspatial.com Official leaderboard
34 gpt-oss-120b 50% high, Terminus-2 Aug 2025 axisspatial.com Official leaderboard
35 Mistral Large 4 50% high, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
36 Gemma 4 26B A4B 46% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
37 Nemotron 3.5 Super VL 46% Reasoning, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
38 glm-4.7-flash 44% Reasoning, Terminus-2 Jan 2026 axisspatial.com Official leaderboard
39 Laguna XS 2.1 44% Reasoning, Terminus-2 Jul 2026 axisspatial.com Official leaderboard
40 Gemma 4 31B IT 42% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
41 MiniMax-M3.1-Flash 41% max, Terminus-2 Oct 2026 axisspatial.com Official leaderboard
42 gpt-oss-20b 32% high, Terminus-2 Aug 2025 axisspatial.com Official leaderboard
43 Mistral Large 3 31% Non-reasoning, Terminus-2 Dec 2025 axisspatial.com Official leaderboard
44 Nemotron 3 Nano Omni 30B-A3B 31% Reasoning, Terminus-2 Apr 2026 axisspatial.com Official leaderboard
45 Nemotron 3.5 Lightning 30B-A3B 16% Reasoning, Terminus-2 Aug 2026 axisspatial.com Official leaderboard
46 Llama 3.3 70B 9.00% Non-reasoning, Terminus-2 Dec 2024 axisspatial.com Official leaderboard
47 Nemotron 3.5 Bolt E4B 5.00% Reasoning, Terminus-2 Oct 2026 axisspatial.com Official leaderboard

The best score, 80%, costs $1.67 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6 Sol at high comes within 5 points of the best for $0.10 a task, 17 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Anthropic OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Opus 5.5, max Claude Opus 5.5, high GPT-6 Sol, high Claude Haiku 5.5, high Claude Haiku 5.5, medium GPT-6 Luna, high GPT-6 Luna, medium 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Opus 5.5, max Claude Haiku 5.5, high

Source: Axis Spatial’s leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 max Claude Code 80% $1.67 48,000
Claude Opus 5.5 high Claude Code 79% $0.18 4,000
GLM-5.3 max Claude Code 77% $0.98 35,000
DeepSeek V4.1 Flash max Claude Code 77% $0.41 94,000
Claude Opus 5.5 xhigh Claude Code 77% $0.40 10,000
Kimi K3 Reasoning Terminus-2 75% $0.41 12,000
Kimi K3 Reasoning Kimi Code 75% $0.56 19,000
DeepSeek V4.1 Flash max Codex 75% $0.13 69,000
GPT-6 Sol high Codex 75% $0.10 4,000
Claude Opus 5 high Claude Code 74% $0.48 8,000
DeepSeek-V4-Flash high Terminus-2 73% $0.13 62,000
Kimi K2.7 Code Reasoning Kimi Code 73% $0.38 24,000
Claude Fable 5.1 medium Claude Code 73% $0.44 4,000
Claude Fable 5.1 high Claude Code 72% $0.71 7,000
Claude Opus 5.5 medium Claude Code 72% $0.14 3,000
GPT-6.1 Sol medium Codex 72% $0.06 2,000
GPT-6.1 Sol high Codex 72% $0.07 3,000
GPT-6.1 Sol xhigh Codex 72% $0.11 5,000
Claude Haiku 5.5 high Claude Code 72% $0.01 8,000
Claude Opus 4.8 high Claude Code 71% $0.44 8,000
GLM-5.2 max Claude Code 71% $0.56 47,000
Claude Sonnet 5.5 high Claude Code 71% $0.08 3,000
DeepSeek-V4-Flash high Codex 70% $0.19 39,000
Kimi K2.7 Code Reasoning Terminus-2 70% $0.25 23,000
GPT-5.6 Sol high Codex 70% $0.26 5,000
Claude Opus 4.7 high Claude Code 69% $0.58 9,000
DeepSeek-V4-Pro max Terminus-2 69% $0.76 46,000
Nemotron 3 Ultra high Terminus-2 69% $0.12 19,000
Qwen 3.8 27B Reasoning Qwen Code 69% $0.32 51,000
GLM-5.3-Flash max Claude Code 69% $0.20 62,000
Kimi K2.6 Reasoning Terminus-2 68% $0.21 27,000
DeepSeek-V4-Pro high Terminus-2 68% $0.57 36,000
MiniMax-M3 Reasoning MiniMax Code 68% $0.18 43,000
GLM-5.2 high Terminus-2 68% $0.39 36,000
GLM-5.2 max Terminus-2 68% $0.48 44,000
Claude Sonnet 5 high Claude Code 68% $0.32 14,000
MiMo-V2.6 Flash Reasoning MiMo Code 68% $0.09 63,000
Claude Sonnet 5.5 xhigh Claude Code 68% $0.16 7,000
Claude Haiku 5.5 xhigh Claude Code 68% $0.02 19,000
Claude Haiku 5.5 medium Claude Code 68% $0.007 5,000
DeepSeek-V4-Flash max Terminus-2 67% $0.11 55,000
GPT-5.6 Luna high Codex 67% $0.02 8,000
GLM-5.3 max Terminus-2 67% $0.83 35,000
GPT-6 Luna xhigh Codex 67% $0.009 9,000
Claude Haiku 5.5 max Claude Code 67% $0.14 75,000
Kimi K2.6 Reasoning Kimi Code 66% $0.60 42,000
GPT-5.6 Luna xhigh Codex 65% $0.03 12,000
Claude Sonnet 5.5 max Claude Code 65% $1.57 90,000
GPT-6 Luna high Codex 63% $0.005 4,000
Qwen 3.6 35B-A3B Reasoning Terminus-2 62% $0.13 37,000
GPT-5.6 Terra high Codex 62% $0.12 4,000
GLM-5.3 low Terminus-2 62% $0.62 19,000
Claude Haiku 5.5 low Claude Code 62% $0.006 4,000
Muse Glimmer high Terminus-2 60% $0.11 19,000
Step 5 Preview Reasoning Terminus-2 60% $0.19 45,000
Qwen 3.6 35B-A3B Reasoning Qwen Code 59% 42,000
Claude Opus 5.5 low Claude Code 59% $0.09 1,000
GLM-5.2 Non-reasoning Terminus-2 58% $0.15 9,000
Claude Sonnet 5.5 medium Claude Code 58% $0.05 2,000
DeepSeek-V4-Flash high Claude Code 56% $0.17 36,000
GPT-5.6 Luna medium Codex 56% $0.01 4,000
GLM-5.3-Flash low Terminus-2 56% $0.13 33,000
Claude Sonnet 5.5 low Claude Code 56% $0.05 1,000
Qwen 3.8 27B xhigh Terminus-2 54% $0.14 29,000
Gemini 3.8 Flash low Terminus-2 54% $0.20 14,000
Gemini 3.1 Flash-Lite high Terminus-2 53% $0.04 23,000
Claude Haiku 4.5 Reasoning Claude Code 51% $0.14 12,000
GLM-5.3-Flash max Terminus-2 51% $0.23 60,000
gpt-oss-120b high Terminus-2 50% $0.04 15,000
GPT-6 Luna medium Codex 50% $0.003 2,000
Mistral Large 4 high Terminus-2 50% $0.12 18,000
gpt-oss-120b high Codex 49% $0.65 67,000
Gemma 4 26B A4B Reasoning Terminus-2 46% $0.03 31,000
Nemotron 3.5 Super VL Reasoning Terminus-2 46% $0.01 11,000
glm-4.7-flash Reasoning Terminus-2 44% $0.05 24,000
Laguna XS 2.1 Reasoning Terminus-2 44% $0.07 33,000
GPT-5.6 Luna low Codex 44% $0.006 2,000
GPT-6 Luna low Codex 44% $0.004 2,000
Gemini 3.1 Flash-Lite low Terminus-2 43% $0.02 11,000
Gemma 4 31B IT Reasoning Terminus-2 42% $0.02 12,000
MiniMax-M3.1-Flash max Terminus-2 41% $0.02 23,000
glm-4.7-flash Reasoning Claude Code 38% $0.87 21,000
Gemini 3.1 Flash-Lite minimal Terminus-2 34% $0.008 3,000
gpt-oss-20b high Terminus-2 32% $0.02 40,000
Mistral Large 3 Non-reasoning Terminus-2 31% $0.05 8,000
Nemotron 3 Nano Omni 30B-A3B Reasoning Terminus-2 31% 20,000
gpt-oss-120b low Terminus-2 28% $0.005 2,000
Nemotron 3.5 Lightning 30B-A3B Reasoning Terminus-2 16% $0.005 7,000
Llama 3.3 70B Non-reasoning Terminus-2 9.00% $0.02 3,000
gpt-oss-20b high Codex 6.00% $0.10 31,000
Nemotron 3.5 Bolt E4B Reasoning Terminus-2 5.00% 24,000

The best score, 80%, costs $1.67 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6 Sol at high comes within 5 points of the best for $0.10 a task, 17 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Anthropic OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Opus 5.5, max Claude Opus 5.5, high GPT-6 Sol, high Claude Haiku 5.5, high Claude Haiku 5.5, medium GPT-6 Luna, high GPT-6 Luna, medium 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Opus 5.5, max Claude Haiku 5.5, high

Source: Axis Spatial’s leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 max Claude Code 80% $1.67 48,000
Claude Opus 5.5 high Claude Code 79% $0.18 4,000
GLM-5.3 max Claude Code 77% $0.98 35,000
DeepSeek V4.1 Flash max Claude Code 77% $0.41 94,000
Claude Opus 5.5 xhigh Claude Code 77% $0.40 10,000
Kimi K3 Reasoning Terminus-2 75% $0.41 12,000
Kimi K3 Reasoning Kimi Code 75% $0.56 19,000
DeepSeek V4.1 Flash max Codex 75% $0.13 69,000
GPT-6 Sol high Codex 75% $0.10 4,000
Claude Opus 5 high Claude Code 74% $0.48 8,000
DeepSeek-V4-Flash high Terminus-2 73% $0.13 62,000
Kimi K2.7 Code Reasoning Kimi Code 73% $0.38 24,000
Claude Fable 5.1 medium Claude Code 73% $0.44 4,000
Claude Fable 5.1 high Claude Code 72% $0.71 7,000
Claude Opus 5.5 medium Claude Code 72% $0.14 3,000
GPT-6.1 Sol medium Codex 72% $0.06 2,000
GPT-6.1 Sol high Codex 72% $0.07 3,000
GPT-6.1 Sol xhigh Codex 72% $0.11 5,000
Claude Haiku 5.5 high Claude Code 72% $0.01 8,000
Claude Opus 4.8 high Claude Code 71% $0.44 8,000
GLM-5.2 max Claude Code 71% $0.56 47,000
Claude Sonnet 5.5 high Claude Code 71% $0.08 3,000
DeepSeek-V4-Flash high Codex 70% $0.19 39,000
Kimi K2.7 Code Reasoning Terminus-2 70% $0.25 23,000
GPT-5.6 Sol high Codex 70% $0.26 5,000
Claude Opus 4.7 high Claude Code 69% $0.58 9,000
DeepSeek-V4-Pro max Terminus-2 69% $0.76 46,000
Nemotron 3 Ultra high Terminus-2 69% $0.12 19,000
Qwen 3.8 27B Reasoning Qwen Code 69% $0.32 51,000
GLM-5.3-Flash max Claude Code 69% $0.20 62,000
Kimi K2.6 Reasoning Terminus-2 68% $0.21 27,000
DeepSeek-V4-Pro high Terminus-2 68% $0.57 36,000
MiniMax-M3 Reasoning MiniMax Code 68% $0.18 43,000
GLM-5.2 high Terminus-2 68% $0.39 36,000
GLM-5.2 max Terminus-2 68% $0.48 44,000
Claude Sonnet 5 high Claude Code 68% $0.32 14,000
MiMo-V2.6 Flash Reasoning MiMo Code 68% $0.09 63,000
Claude Sonnet 5.5 xhigh Claude Code 68% $0.16 7,000
Claude Haiku 5.5 xhigh Claude Code 68% $0.02 19,000
Claude Haiku 5.5 medium Claude Code 68% $0.007 5,000
DeepSeek-V4-Flash max Terminus-2 67% $0.11 55,000
GPT-5.6 Luna high Codex 67% $0.02 8,000
GLM-5.3 max Terminus-2 67% $0.83 35,000
GPT-6 Luna xhigh Codex 67% $0.009 9,000
Claude Haiku 5.5 max Claude Code 67% $0.14 75,000
Kimi K2.6 Reasoning Kimi Code 66% $0.60 42,000
GPT-5.6 Luna xhigh Codex 65% $0.03 12,000
Claude Sonnet 5.5 max Claude Code 65% $1.57 90,000
GPT-6 Luna high Codex 63% $0.005 4,000
Qwen 3.6 35B-A3B Reasoning Terminus-2 62% $0.13 37,000
GPT-5.6 Terra high Codex 62% $0.12 4,000
GLM-5.3 low Terminus-2 62% $0.62 19,000
Claude Haiku 5.5 low Claude Code 62% $0.006 4,000
Muse Glimmer high Terminus-2 60% $0.11 19,000
Step 5 Preview Reasoning Terminus-2 60% $0.19 45,000
Qwen 3.6 35B-A3B Reasoning Qwen Code 59% 42,000
Claude Opus 5.5 low Claude Code 59% $0.09 1,000
GLM-5.2 Non-reasoning Terminus-2 58% $0.15 9,000
Claude Sonnet 5.5 medium Claude Code 58% $0.05 2,000
DeepSeek-V4-Flash high Claude Code 56% $0.17 36,000
GPT-5.6 Luna medium Codex 56% $0.01 4,000
GLM-5.3-Flash low Terminus-2 56% $0.13 33,000
Claude Sonnet 5.5 low Claude Code 56% $0.05 1,000
Qwen 3.8 27B xhigh Terminus-2 54% $0.14 29,000
Gemini 3.8 Flash low Terminus-2 54% $0.20 14,000
Gemini 3.1 Flash-Lite high Terminus-2 53% $0.04 23,000
Claude Haiku 4.5 Reasoning Claude Code 51% $0.14 12,000
GLM-5.3-Flash max Terminus-2 51% $0.23 60,000
gpt-oss-120b high Terminus-2 50% $0.04 15,000
GPT-6 Luna medium Codex 50% $0.003 2,000
Mistral Large 4 high Terminus-2 50% $0.12 18,000
gpt-oss-120b high Codex 49% $0.65 67,000
Gemma 4 26B A4B Reasoning Terminus-2 46% $0.03 31,000
Nemotron 3.5 Super VL Reasoning Terminus-2 46% $0.01 11,000
glm-4.7-flash Reasoning Terminus-2 44% $0.05 24,000
Laguna XS 2.1 Reasoning Terminus-2 44% $0.07 33,000
GPT-5.6 Luna low Codex 44% $0.006 2,000
GPT-6 Luna low Codex 44% $0.004 2,000
Gemini 3.1 Flash-Lite low Terminus-2 43% $0.02 11,000
Gemma 4 31B IT Reasoning Terminus-2 42% $0.02 12,000
MiniMax-M3.1-Flash max Terminus-2 41% $0.02 23,000
glm-4.7-flash Reasoning Claude Code 38% $0.87 21,000
Gemini 3.1 Flash-Lite minimal Terminus-2 34% $0.008 3,000
gpt-oss-20b high Terminus-2 32% $0.02 40,000
Mistral Large 3 Non-reasoning Terminus-2 31% $0.05 8,000
Nemotron 3 Nano Omni 30B-A3B Reasoning Terminus-2 31% 20,000
gpt-oss-120b low Terminus-2 28% $0.005 2,000
Nemotron 3.5 Lightning 30B-A3B Reasoning Terminus-2 16% $0.005 7,000
Llama 3.3 70B Non-reasoning Terminus-2 9.00% $0.02 3,000
gpt-oss-20b high Codex 6.00% $0.10 31,000
Nemotron 3.5 Bolt E4B Reasoning Terminus-2 5.00% 24,000

Number one over time

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

Anthropic
Axis Spatial leaderboard 1 leader

Source: Axis Spatial’s leaderboard

Scores over time

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

Plot with other benchmarks
On Axis Spatial’s leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Kimi K3 Claude Opus 4.7 Qwen 3.6 35B-A3B Gemini 3.1 Flash-Lite Claude Haiku 4.5 Llama 3.3 70B 0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Claude Opus 5.5 DeepSeek-V4-Flash

Source: Axis Spatial’s leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

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.

OpenAI Anthropic
Earth-Bench V2 2 leaders
Earth-Bench Not tested since Sep 2025
GPT-5 66%

Source: Official leaderboards

More in Tool use

All 265 in Tool use
  • AutomationBench In use Agent completes end-to-end business workflows inside a simulated company exposing dozens of REST APIs. Graded on resulting state. Gemini 4 Neon 56.3%
  • MCP Atlas In use Multi-step workflows over real MCP servers. Scores tool-use competency on realistic tasks. Gemini 4 Argon 90.7%
  • BFCL V4 In use Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. Qwen3.7-Max 75%
  • τ³-bench Banking In use Agents act as banking customer-service reps over many turns with a simulated user, following policy. Pass^1 on the final database state. Qwen 3.8 Max 55.1%
  • Toolathlon-Verified In use 108 long-horizon tasks across 32 apps through about 600 MCP tools. Execution-based checkers inspect final environment state. GLM-5.3-Flash 78.4%
  • DeepSearchQA In use Questions that need multi-step search and synthesis. Final answers graded against references. Kimi K3 95%

Data audit

All 11 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (48 entries).
  • Pass Models scored 47 models we track, plus 1 other entry on its leaderboard.
  • Pass Newest score The newest is from October 2026, for Nemotron 3.5 Bolt E4B.
  • Pass Frontier models 4 of the top 10 have a score, led by Claude Opus 5.5 (#5).
  • Pass Headroom The best is 80%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on Axis Spatial: 91 rows for models we track and 1 other entry at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check