Which model should run your agents?
The Shogo Agent Index ranks models only on agentic work: fixing real repos, staying on task for hours, producing professional deliverables and driving a computer. 17 benchmarks, weighted, every score sourced, with the price an agent actually pays.
2,941 scores · 1,266 endpoints on 83 hosts
Top 25 models for agents
All 92 ranked models →| Model | Agent Index | Coding | Long-hzn | Know. work | Comp. use | Sec/ctx | Measured | Agent $/M | Context |
|---|---|---|---|---|---|---|---|---|---|
| Claude Fable 5.2 AnthropicSupported | 76.4 71.7–81.2 | 86 | 78 | 67 | 77 | 65 | 32% | $3.21 | 1M |
| Gemini 4 Carbon GoogleSupported | 73.6 67.2–80 | 82 | 80 | 63 | 68 | 63 | 27% | $0.84 | 1M |
| Claude Opus 5.5 AnthropicSupported | 70.9 64.8–76.9 | 81 | 76 | 58 | 66 | 62 | 32% | $1.69 | 1M |
| GPT-6 Astra OpenAISupported | 70.9 66.5–75.4 | 80 | 80 | 56 | 67 | 63 | 38% | $4.62 | 1.1M |
| GPT-6.1 Sol OpenAIEstimated | 70.0 62.2–77.7 | 80 | 78 | 54 | 66 | 62 | 18% | $0.84 | 1.1M |
| Gemini 4 Argon GoogleSupported | 69.5 62.9–76.2 | 77 | 78 | 59 | 61 | 61 | 32% | $0.84 | 1M |
| Gemini 4 Hydrogen GoogleEstimated | 69.4 61.8–77.1 | 78 | 78 | 58 | 61 | 61 | 23% | $0.63 | 1M |
| GPT-6.1 Terra OpenAIEstimated | 69.0 60.5–77.5 | 79 | 74 | 56 | 63 | 62 | 14% | $0.42 | 1.1M |
| Claude Sonnet 5.5 AnthropicEstimated | 69.0 61.3–76.7 | 79 | 75 | 57 | 61 | 61 | 23% | $0.84 | 1M |
| Claude Fable 5.1 AnthropicSupported | 68.2 62.7–73.7 | 80 | 71 | 55 | 62 | 61 | 36% | $4.01 | 1M |
| Claude Opus 5 AnthropicSupported | 65.4 59.8–71 | 74 | 74 | 52 | 56 | 59 | 39% | $2.31 | 1M |
| Claude Fable 5 AnthropicSupported | 64.1 59–69.3 | 74 | 69 | 52 | 55 | 59 | 38% | $4.62 | 1M |
| GPT-6 Sol OpenAIEstimated | 63.6 54.2–73 | 72 | 76 | 48 | 54 | 59 | 14% | $0.92 | 1.1M |
| GPT-5.6 Sol OpenAISupported | 61.8 57.1–66.5 | 71 | 72 | 46 | 53 | 58 | 43% | $1.85 | 1.1M |
| Muse Wonder MetaEstimated | 60.9 51.9–69.9 | 68 | 70 | 50 | 50 | 57 | 22% | $1.69 | 1M |
| GPT-5.6 Terra OpenAIEstimated | 58.8 50.7–66.9 | 66 | 69 | 47 | 48 | 56 | 24% | $1.02 | 1.1M |
| DeepSeek V4.1 Flash 1017 DeepSeek· openEstimated | 57.6 44.4–70.8 | 63 | 66 | 49 | 45 | 54 | 12% | $0.050 | 1M |
| Claude Haiku 5.5 AnthropicEstimated | 57.5 48.4–66.5 | 67 | 62 | 46 | 47 | 55 | 23% | $0.050 | 1M |
| GPT-5.5 OpenAISupported | 56.7 52.5–61 | 63 | 67 | 46 | 43 | 55 | 60% | $2.55 | 1.1M |
| Claude Opus 4.8 AnthropicSupported | 56.3 50.9–61.8 | 63 | 66 | 43 | 47 | 55 | 44% | $2.31 | 1M |
| Gemini 3.8 Flash GoogleSupported | 56.2 48.7–63.7 | 63 | 64 | 47 | 44 | 53 | 34% | $0.35 | 1M |
| Grok 4.6 xAISupported | 56.2 48.3–64 | 61 | 68 | 47 | 42 | 52 | 29% | $0.98 | 500K |
| Gemini 3.7 Flash GoogleEstimated | 55.7 44.2–67.1 | 61 | 64 | 48 | 43 | 53 | 24% | $0.35 | 1M |
| Inkling 2 Small thinkingmachines· openEstimated | 55.2 42.3–68.2 | 62 | 66 | 42 | 44 | 54 | 13% | $0.14 | 1M |
| GLM-5.3 Z.ai (Zhipu)· openSupported | 55.2 47.7–62.7 | 62 | 68 | 43 | 40 | 51 | 33% | $0.61 | 1M |
Tick models to compare them side by side. Pillar scores in white include measured results; grey ones are fully estimated. Shaded band = 95% range.
Agent performance per dollar
Agents resend a long, mostly cached context every turn. We price each model at 20 input tokens per output token with 85% cache hits, which is closer to an agent’s bill than the usual 3:1 blend.
The five pillars
- Agentic coding35%
Fixing real repos, working in a terminal and shipping multi-file changes without a human in the loop.
- Long-horizon autonomy20%
Staying on task for hours: long software tasks, running a business, training a model end to end.
- Knowledge-work agents25%
Professional deliverables graded by experts: banking, consulting, law, research reports, paid freelance jobs.
- Computer use10%
Operating a real desktop and browser through screenshots, mouse and keyboard.
- Security & context learning10%
Capture-the-flag security tasks and learning new rules from long context on the fly.
Put the winner to work
Shogo runs your agents on whichever model fits the job, and routes each step to the cheapest model that can do it. Same models as this leaderboard, no API keys to juggle.
Build an agent in ShogoLatest changes
- Launched Shogo Agent Index v1.0: 92 ranked models, 564 models tracked, 46 benchmarks.