Coding Agents
Compare model + harness configurations measured by Artificial Analysis: score, cost and time per task.
Coding Agent IndexDeepSWE v1.1SWE-Atlas-QnATerminal-Bench v4
- Configurations
- 40
- Harnesses
- 10
- Best index
- 68.4
- Claude Code · Sonnet 5.5 (max)
- Cheapest in the top 10
- $1.04
- Codex · GPT-6.1 Sol (xhigh)
Score and cost
Each line joins the effort levels of one configuration; the large point marks the best one. Top right: the best score for the lowest cost. Hover or move through the points with the keyboard to read their values.
Models shown27 of 27
Select a model to show or hide it; hovering it brings it forward on the chart. The axes fit the models shown.
40 / 40 configurations
- #168.4
Claude Code
Sonnet 5.5max- DeepSWE v1.1
- 72.0%
- SWE-Atlas-QnA
- 66.9%
- Terminal-Bench v4
- 66.2%
- Cost
- $14.19
- Time
- 1 h 27
- Output tokens
- 601K
- #268.1
Devin Fusion CLI
Claude Opus 5.5maxGPT-6 Lunahigh- DeepSWE v1.1
- 72.6%
- SWE-Atlas-QnA
- 69.6%
- Terminal-Bench v4
- 62.1%
- Cost
- $9.01
- Time
- 54 min 50 s
- Output tokens
- 296K
- #368.0
Muse Code
Muse Wonder Fullmax- DeepSWE v1.1
- 77.0%
- SWE-Atlas-QnA
- 68.0%
- Terminal-Bench v4
- 59.1%
- Cost
- —
- Time
- 28 min 46 s
- Output tokens
- 85.1K
- #466.0
Claude Code
Opus 5.5max- DeepSWE v1.1
- 68.4%
- SWE-Atlas-QnA
- 66.4%
- Terminal-Bench v4
- 63.1%
- Cost
- $13.04
- Time
- 1 h 04
- Output tokens
- 333K
- #563.8
Antigravity CLI
Gemini 4 Argon- DeepSWE v1.1
- 78.8%
- SWE-Atlas-QnA
- 56.5%
- Terminal-Bench v4
- 56.1%
- Cost
- $5.84
- Time
- 34 min 30 s
- Output tokens
- 112K
- #663.5
Muse Code
Muse Wondermax- DeepSWE v1.1
- 75.8%
- SWE-Atlas-QnA
- 67.2%
- Terminal-Bench v4
- 47.5%
- Cost
- $8.05
- Time
- 22 min 27 s
- Output tokens
- 90K
- #762.9
Codex
GPT-6.1 Solxhigh- DeepSWE v1.1
- 73.2%
- SWE-Atlas-QnA
- 61.0%
- Terminal-Bench v4
- 54.5%
- Cost
- $1.04
- Time
- 15 min 31 s
- Output tokens
- 35.4K
- #862.9
Claude Code
Sonnet 5.5xhigh- DeepSWE v1.1
- 68.4%
- SWE-Atlas-QnA
- 62.1%
- Terminal-Bench v4
- 58.1%
- Cost
- $3.33
- Time
- 26 min 59 s
- Output tokens
- 126K
- #962.2
Claude Code
Fable 5.1 (with fallback)max- DeepSWE v1.1
- 64.3%
- SWE-Atlas-QnA
- 64.8%
- Terminal-Bench v4
- 57.6%
- Cost
- $12.39
- Time
- 34 min 50 s
- Output tokens
- 134K
- #1061.7
Devin Fusion CLI
Claude Fable 5.1xhighSWE-2medium- DeepSWE v1.1
- 63.1%
- SWE-Atlas-QnA
- 65.9%
- Terminal-Bench v4
- 56.1%
- Cost
- $7.90
- Time
- 35 min 47 s
- Output tokens
- 101K
- #1161.6
Codex
GPT-6 Astramax- DeepSWE v1.1
- 67.6%
- SWE-Atlas-QnA
- 61.8%
- Terminal-Bench v4
- 55.6%
- Cost
- $7.47
- Time
- 29 min 22 s
- Output tokens
- 47.1K
- #1261.4
Codex
GPT-6.1 Solmedium- DeepSWE v1.1
- 72.0%
- SWE-Atlas-QnA
- 60.8%
- Terminal-Bench v4
- 51.5%
- Cost
- $0.705
- Time
- 10 min 51 s
- Output tokens
- 22.4K
- #1360.1
Codex
GPT-6.1 Solhigh- DeepSWE v1.1
- 70.5%
- SWE-Atlas-QnA
- 59.9%
- Terminal-Bench v4
- 50.0%
- Cost
- $0.889
- Time
- 13 min 20 s
- Output tokens
- 28.7K
- #1460.1
Codex
GPT-6.1 Solmax- DeepSWE v1.1
- 69.6%
- SWE-Atlas-QnA
- 57.8%
- Terminal-Bench v4
- 53.0%
- Cost
- $1.55
- Time
- 24 min 23 s
- Output tokens
- 61.7K
- #1559.7
Claude Code
Opus 5max- DeepSWE v1.1
- 62.5%
- SWE-Atlas-QnA
- 62.1%
- Terminal-Bench v4
- 54.5%
- Cost
- $10.79
- Time
- 41 min 56 s
- Output tokens
- 137K
- #1658.9
Devin Fusion CLI
GPT-6 AstraxhighSWE-2medium- DeepSWE v1.1
- 67.3%
- SWE-Atlas-QnA
- 59.4%
- Terminal-Bench v4
- 50.0%
- Cost
- $4.54
- Time
- 24 min 42 s
- Output tokens
- 56.6K
- #1757.2
Codex
GPT-6.1 Sollow- DeepSWE v1.1
- 67.6%
- SWE-Atlas-QnA
- 55.1%
- Terminal-Bench v4
- 49.0%
- Cost
- $0.499
- Time
- 8 min 38 s
- Output tokens
- 14.7K
- #1856.7
Codex
GPT-6 Solmax- DeepSWE v1.1
- 69.0%
- SWE-Atlas-QnA
- 57.5%
- Terminal-Bench v4
- 43.4%
- Cost
- $2.99
- Time
- 22 min 18 s
- Output tokens
- 62.2K
- #1956.3
Grok Build
Grok 4.7xhigh- DeepSWE v1.1
- 72.6%
- SWE-Atlas-QnA
- 62.9%
- Terminal-Bench v4
- 33.3%
- Cost
- $8.82
- Time
- 39 min 14 s
- Output tokens
- 123K
- #2055.0
Claude Code
Sonnet 5.5high- DeepSWE v1.1
- 66.7%
- SWE-Atlas-QnA
- 56.5%
- Terminal-Bench v4
- 41.9%
- Cost
- $1.24
- Time
- 12 min 15 s
- Output tokens
- 41.5K
- #2154.6
Codex
GPT-5.6 Solmax- DeepSWE v1.1
- 72.3%
- SWE-Atlas-QnA
- 54.0%
- Terminal-Bench v4
- 37.4%
- Cost
- $6.35
- Time
- 20 min 38 s
- Output tokens
- 49.2K
- #2254.3
Muse Code
Muse Spark 1.3max- DeepSWE v1.1
- 71.7%
- SWE-Atlas-QnA
- 59.4%
- Terminal-Bench v4
- 31.8%
- Cost
- $3.98
- Time
- 18 min 27 s
- Output tokens
- 263K
- #2353.6
Opencode
GLM-5.3- DeepSWE v1.1
- 61.4%
- SWE-Atlas-QnA
- 59.4%
- Terminal-Bench v4
- 39.9%
- Cost
- $4.24
- Time
- 48 min 08 s
- Output tokens
- 74.3K
- #2451.9
Kimi Code CLI
Kimi K3- DeepSWE v1.1
- 68.4%
- SWE-Atlas-QnA
- 66.1%
- Terminal-Bench v4
- 21.2%
- Cost
- $5.05
- Time
- 1 h 01
- Output tokens
- 93.2K
- #2551.2
Claude Code
Step 5 Preview- DeepSWE v1.1
- 61.9%
- SWE-Atlas-QnA
- 58.3%
- Terminal-Bench v4
- 33.3%
- Cost
- $1.66
- Time
- 56 min 21 s
- Output tokens
- 126K
- #2648.3
Muse Code 1.0.2 RC
Muse Spark 1.3xhigh- DeepSWE v1.1
- 73.2%
- SWE-Atlas-QnA
- 54.6%
- Terminal-Bench v4
- 17.2%
- Cost
- $3.47
- Time
- 18 min 18 s
- Output tokens
- 146K
- #2747.0
Grok Build
Grok 4.6xhigh- DeepSWE v1.1
- 64.9%
- SWE-Atlas-QnA
- 58.3%
- Terminal-Bench v4
- 17.7%
- Cost
- $3.57
- Time
- 19 min 27 s
- Output tokens
- 56K
- #2845.9
Claude Code
Sonnet 5.5medium- DeepSWE v1.1
- 65.5%
- SWE-Atlas-QnA
- 44.9%
- Terminal-Bench v4
- 27.3%
- Cost
- $0.619
- Time
- 8 min 31 s
- Output tokens
- 20.5K
- #2943.3
Claude Code
Qwen3.8max- DeepSWE v1.1
- 51.0%
- SWE-Atlas-QnA
- 62.1%
- Terminal-Bench v4
- 16.7%
- Cost
- $3.48
- Time
- 1 h 03
- Output tokens
- 106K
- #3043.2
Codex
GPT-5.6 Lunamax- DeepSWE v1.1
- 66.4%
- SWE-Atlas-QnA
- 48.7%
- Terminal-Bench v4
- 14.6%
- Cost
- $0.438
- Time
- 23 min 32 s
- Output tokens
- 71.3K
- #3143.1
Codex
DeepSeek V4 Pro 0813max- DeepSWE v1.1
- 57.2%
- SWE-Atlas-QnA
- 61.8%
- Terminal-Bench v4
- 10.1%
- Cost
- $0.238
- Time
- 40 min 18 s
- Output tokens
- 116K
- #3242.1
Claude Code
Sonnet 5.5low- DeepSWE v1.1
- 61.9%
- SWE-Atlas-QnA
- 39.0%
- Terminal-Bench v4
- 25.3%
- Cost
- $0.483
- Time
- 6 min 21 s
- Output tokens
- 15.8K
- #3341.9
Antigravity SDK v0.1.12
Gemini 3.8 Flashhigh- DeepSWE v1.1
- 65.8%
- SWE-Atlas-QnA
- 45.2%
- Terminal-Bench v4
- 14.6%
- Cost
- $2.47
- Time
- 11 min 43 s
- Output tokens
- 116K
- #3441.4
Claude Code
Haiku 5.5xhigh- DeepSWE v1.1
- 49.0%
- SWE-Atlas-QnA
- 51.6%
- Terminal-Bench v4
- 23.7%
- Cost
- $0.61
- Time
- 25 min 53 s
- Output tokens
- 190K
- #3541.1
Codex
GPT-6 Lunamax- DeepSWE v1.1
- 63.7%
- SWE-Atlas-QnA
- 44.4%
- Terminal-Bench v4
- 15.2%
- Cost
- $0.176
- Time
- 21 min 26 s
- Output tokens
- 98.3K
- #3638.7
Codex
DeepSeek V4 Flash 0731max- DeepSWE v1.1
- 54.3%
- SWE-Atlas-QnA
- 51.3%
- Terminal-Bench v4
- 10.6%
- Cost
- $0.0851
- Time
- 18 min 17 s
- Output tokens
- 80.4K
- #3736.5
Claude Code
Haiku 5.5max- DeepSWE v1.1
- 46.0%
- SWE-Atlas-QnA
- 33.6%
- Terminal-Bench v4
- 29.8%
- Cost
- $2.58
- Time
- 1 h 27
- Output tokens
- 990K
- #3835.1
Claude Code
Haiku 5.5high- DeepSWE v1.1
- 46.3%
- SWE-Atlas-QnA
- 46.2%
- Terminal-Bench v4
- 12.6%
- Cost
- $0.365
- Time
- 18 min 27 s
- Output tokens
- 105K
- #3933.6
Claude Code
Haiku 5.5medium- DeepSWE v1.1
- 47.5%
- SWE-Atlas-QnA
- 41.7%
- Terminal-Bench v4
- 11.6%
- Cost
- $0.226
- Time
- 13 min 57 s
- Output tokens
- 67.6K
- #4028.2
Claude Code
Haiku 5.5low- DeepSWE v1.1
- 45.4%
- SWE-Atlas-QnA
- 33.1%
- Terminal-Bench v4
- 6.1%
- Cost
- $0.143
- Time
- 11 min 34 s
- Output tokens
- 48.1K
| # | Harness | Model | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Claude Code | Sonnet 5.5max | 68.4 | 72.0% | 66.9% | 66.2% | $14.19 | 1 h 27 | 601K | |
| 2 | Devin Fusion CLI | Claude Opus 5.5maxGPT-6 Lunahigh | 68.1 | 72.6% | 69.6% | 62.1% | $9.01 | 54 min 50 s | 296K |
| 3 | Muse Code | Muse Wonder Fullmax | 68.0 | 77.0% | 68.0% | 59.1% | — | 28 min 46 s | 85.1K |
| 4 | Claude Code | Opus 5.5max | 66.0 | 68.4% | 66.4% | 63.1% | $13.04 | 1 h 04 | 333K |
| 5 | Antigravity CLI | Gemini 4 Argon | 63.8 | 78.8% | 56.5% | 56.1% | $5.84 | 34 min 30 s | 112K |
| 6 | Muse Code | Muse Wondermax | 63.5 | 75.8% | 67.2% | 47.5% | $8.05 | 22 min 27 s | 90K |
| 7 | Codex | GPT-6.1 Solxhigh | 62.9 | 73.2% | 61.0% | 54.5% | $1.04 | 15 min 31 s | 35.4K |
| 8 | Claude Code | Sonnet 5.5xhigh | 62.9 | 68.4% | 62.1% | 58.1% | $3.33 | 26 min 59 s | 126K |
| 9 | Claude Code | Fable 5.1 (with fallback)max | 62.2 | 64.3% | 64.8% | 57.6% | $12.39 | 34 min 50 s | 134K |
| 10 | Devin Fusion CLI | Claude Fable 5.1xhighSWE-2medium | 61.7 | 63.1% | 65.9% | 56.1% | $7.90 | 35 min 47 s | 101K |
| 11 | Codex | GPT-6 Astramax | 61.6 | 67.6% | 61.8% | 55.6% | $7.47 | 29 min 22 s | 47.1K |
| 12 | Codex | GPT-6.1 Solmedium | 61.4 | 72.0% | 60.8% | 51.5% | $0.705 | 10 min 51 s | 22.4K |
| 13 | Codex | GPT-6.1 Solhigh | 60.1 | 70.5% | 59.9% | 50.0% | $0.889 | 13 min 20 s | 28.7K |
| 14 | Codex | GPT-6.1 Solmax | 60.1 | 69.6% | 57.8% | 53.0% | $1.55 | 24 min 23 s | 61.7K |
| 15 | Claude Code | Opus 5max | 59.7 | 62.5% | 62.1% | 54.5% | $10.79 | 41 min 56 s | 137K |
| 16 | Devin Fusion CLI | GPT-6 AstraxhighSWE-2medium | 58.9 | 67.3% | 59.4% | 50.0% | $4.54 | 24 min 42 s | 56.6K |
| 17 | Codex | GPT-6.1 Sollow | 57.2 | 67.6% | 55.1% | 49.0% | $0.499 | 8 min 38 s | 14.7K |
| 18 | Codex | GPT-6 Solmax | 56.7 | 69.0% | 57.5% | 43.4% | $2.99 | 22 min 18 s | 62.2K |
| 19 | Grok Build | Grok 4.7xhigh | 56.3 | 72.6% | 62.9% | 33.3% | $8.82 | 39 min 14 s | 123K |
| 20 | Claude Code | Sonnet 5.5high | 55.0 | 66.7% | 56.5% | 41.9% | $1.24 | 12 min 15 s | 41.5K |
| 21 | Codex | GPT-5.6 Solmax | 54.6 | 72.3% | 54.0% | 37.4% | $6.35 | 20 min 38 s | 49.2K |
| 22 | Muse Code | Muse Spark 1.3max | 54.3 | 71.7% | 59.4% | 31.8% | $3.98 | 18 min 27 s | 263K |
| 23 | Opencode | GLM-5.3 | 53.6 | 61.4% | 59.4% | 39.9% | $4.24 | 48 min 08 s | 74.3K |
| 24 | Kimi Code CLI | Kimi K3 | 51.9 | 68.4% | 66.1% | 21.2% | $5.05 | 1 h 01 | 93.2K |
| 25 | Claude Code | Step 5 Preview | 51.2 | 61.9% | 58.3% | 33.3% | $1.66 | 56 min 21 s | 126K |
| 26 | Muse Code 1.0.2 RC | Muse Spark 1.3xhigh | 48.3 | 73.2% | 54.6% | 17.2% | $3.47 | 18 min 18 s | 146K |
| 27 | Grok Build | Grok 4.6xhigh | 47.0 | 64.9% | 58.3% | 17.7% | $3.57 | 19 min 27 s | 56K |
| 28 | Claude Code | Sonnet 5.5medium | 45.9 | 65.5% | 44.9% | 27.3% | $0.619 | 8 min 31 s | 20.5K |
| 29 | Claude Code | Qwen3.8max | 43.3 | 51.0% | 62.1% | 16.7% | $3.48 | 1 h 03 | 106K |
| 30 | Codex | GPT-5.6 Lunamax | 43.2 | 66.4% | 48.7% | 14.6% | $0.438 | 23 min 32 s | 71.3K |
| 31 | Codex | DeepSeek V4 Pro 0813max | 43.1 | 57.2% | 61.8% | 10.1% | $0.238 | 40 min 18 s | 116K |
| 32 | Claude Code | Sonnet 5.5low | 42.1 | 61.9% | 39.0% | 25.3% | $0.483 | 6 min 21 s | 15.8K |
| 33 | Antigravity SDK v0.1.12 | Gemini 3.8 Flashhigh | 41.9 | 65.8% | 45.2% | 14.6% | $2.47 | 11 min 43 s | 116K |
| 34 | Claude Code | Haiku 5.5xhigh | 41.4 | 49.0% | 51.6% | 23.7% | $0.61 | 25 min 53 s | 190K |
| 35 | Codex | GPT-6 Lunamax | 41.1 | 63.7% | 44.4% | 15.2% | $0.176 | 21 min 26 s | 98.3K |
| 36 | Codex | DeepSeek V4 Flash 0731max | 38.7 | 54.3% | 51.3% | 10.6% | $0.0851 | 18 min 17 s | 80.4K |
| 37 | Claude Code | Haiku 5.5max | 36.5 | 46.0% | 33.6% | 29.8% | $2.58 | 1 h 27 | 990K |
| 38 | Claude Code | Haiku 5.5high | 35.1 | 46.3% | 46.2% | 12.6% | $0.365 | 18 min 27 s | 105K |
| 39 | Claude Code | Haiku 5.5medium | 33.6 | 47.5% | 41.7% | 11.6% | $0.226 | 13 min 57 s | 67.6K |
| 40 | Claude Code | Haiku 5.5low | 28.2 | 45.4% | 33.1% | 6.1% | $0.143 | 11 min 34 s | 48.1K |
Data from Artificial Analysis · checked every 6 hours. Filtering preserves the index rank.