DeepSeek-V3
Open DeepSeek MoE chat model for coding, math, and general reasoning
Shogo Agent Index
UnrankedNot ranked yet: 1 agentic benchmark measured, and we need at least 2 plus an overall capability score. It will be ranked as results are published.
Agentic coding
14
estimated
Long-horizon autonomy
26
50% measured
Knowledge-work agents
14
estimated
Computer use
4
estimated
Security & context learning
13
estimated
Price
- Input
- $0.29/M
- Output
- $1.16/M
- Cache read
- —/M
- 3:1 blend
- $0.51/M
- Agent-effective
- $0.33/M
Agent-effective = 20 input tokens per output token, 85% cache hits. Why.
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 45 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 5 | |
| DeepSWE | est. | 6% | predicted 7 | |
| FrontierSWE | est. | 5% | predicted 1 | |
| CursorBench | est. | 4% | predicted 15 | |
| GSO-Bench | est. | 4% | predicted 1 | |
| METR Time Horizons | 47.4% | 10% | source | |
| Vending-Bench 2 | est. | 6% | predicted 1 | |
| PostTrainBench | est. | 4% | predicted 10 | |
| APEX-Agents | est. | 8% | predicted 8 | |
| GDPval | est. | 7% | predicted 14 | |
| Remote Labor Index | est. | 5% | predicted 0 | |
| DeepResearch Bench | est. | 5% | predicted 40 | |
| OSWorld 2.0 | est. | 6% | predicted 0 | |
| OSWorld | est. | 4% | predicted 10 | |
| Cybench | est. | 5% | predicted 14 | |
| CL-bench | est. | 5% | predicted 11 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| Aider Polyglot | Coding | 48.4% | source |
| CritPt | Reasoning | 0.0% | source |
| GPQA Diamond | Reasoning | 56.5% | source |
| MATH Level 5 | Math | 64.9% | source |
| MMLU | Legacy | 87.2% | source |
| OTIS Mock AIME | Math | 15.8% | source |
| SciCode | Coding | 35.8% | source |
| SimpleBench | Reasoning | 18.9% | source |