Loading recommendations for the selected task…
Start with the job, inspect the evidence, then price your own workload. Recommendations compare models within one benchmark source and evaluation setup. They cover the evidence we have, not every model on the market.
Multi-step logic, planning, or analysis. Quality measured by SimpleBench — adversarial reasoning questions engineered to stay hard for frontier LLMs. Cost is blended input+output since reasoning often uses long context.
Quality threshold: 10.0% on simple-bench. Ranking cost uses a 70% input / 30% output blend. Your workload estimate below uses your own token counts.
Comparison group: simple-bench · simple-bench · score=avg@5. 16 evidence groups tracked; 25 records excluded from this recommendation because of comparison coverage, pricing, or the quality threshold.
Rank 1 in this comparison · Anthropic
Anthropic
simple-bench · score=AVG@5
Source: simple-bench · observed 2026-09-25
Provider selected by the ranking’s cost measure. Review its exact listing and workload price before buying.
Inspect model evidence Fill up your tankRank 2 in this comparison · Anthropic
Anthropic
simple-bench · score=AVG@5
Source: simple-bench · observed 2026-09-23
Provider selected by the ranking’s cost measure. Review its exact listing and workload price before buying.
Inspect model evidence Fill up your tankRank 3 in this comparison · OpenAI
azure
simple-bench · score=AVG@5
Source: simple-bench · observed 2026-09-23
Provider selected by the ranking’s cost measure. Review its exact listing and workload price before buying.
Inspect model evidenceVerified purchase destination unavailable for this provider.
Compare exact models and providers before you fill up.
Anthropic
Loading estimate…
Anthropic
Loading estimate…
Estimates use tracked list rates. They are not actual charges or a billing commitment. Taxes, tool fees, negotiated rates and provider-specific limits may differ. Provider sign-in is required to buy credits.
| Model / provider | Benchmark result | Input / 1M | Output / 1M |
|---|---|---|---|
| Gemini 3.8 Flash [score=AVG@5] | 82.4% | $0.75 | $3.75 |
| Muse Spark 1.3 [score=AVG@5] meta | 81.8% | $1.25 | $4.25 |
| Claude Opus 5 [score=AVG@5] Anthropic | 80.6% | $5 | $25 |
| Gemini 3.1 Pro Preview [score=AVG@5] | 79.6% | $2 | $12 |
| Gemini 3.5 Flash [score=AVG@5] | 76.7% | $1.5 | $9 |
| Gemini 3 Pro Preview [score=AVG@5] vertex_ai-language-models | 76.4% | $2 | $12 |
| Grok 4.6 [score=AVG@5] xai | 75.9% | $2 | $6 |
| Muse Spark 1.2 [score=AVG@5] meta | 74.5% | $1.25 | $4.25 |
| GPT-6 Sol [score=AVG@5] azure | 73.1% | $2 | $10 |
| Grok 4.5 [score=AVG@5] xai | 70.0% | $2 | $6 |
| GPT-5.5 [score=AVG@5] azure | 69.0% | $5 | $30 |
| Claude Opus 4.6 [score=AVG@5] |
Need raw data for an integration? Read the developer API documentation.
Anthropic
| 67.6% |
| $5 |
| $25 |
| DeepSeek V4.1 Flash [score=AVG@5] baidu | 66.7% | $0.167832 | $0.671329 |
| GLM 5.3 [score=AVG@5] baidu | 66.2% | $0.671329 | $2.34965 |
| Claude Opus 4.8 [score=AVG@5] Anthropic | 64.8% | $5 | $25 |
| Claude Opus 4.5 [score=AVG@5] Anthropic | 62.0% | $5 | $25 |
| Claude Opus 4.7 [score=AVG@5] Anthropic | 61.7% | $5 | $25 |
| GPT-5 Pro [score=AVG@5] azure | 61.6% | $15 | $120 |
| DeepSeek V4 Flash [score=AVG@5] venice | 61.1% | $0.138 | $0.275 |
| Gemini 3 Flash Preview [score=AVG@5] | 61.1% | $0.5 | $3 |
| Claude Sonnet 5 [score=AVG@5] Anthropic | 60.6% | $2 | $10 |
| Grok 4 [score=AVG@5] azure_ai | 60.5% | $3 | $15 |
All contenders shown belong to the comparison group above. Benchmark settings and price observations can change; inspect source evidence before relying on a result.