Internal reports suggest Google's flagship AI, Gemini, is facing significant delays because the underlying technology has failed to meet the company's high performance benchmarks.
The leading AI models for real-world work
Updated July 17, 2026
| Rank | Model & Provider | Overall Score | SWE-bench Verified | API Price / 1M (in / out) | Weekly Change |
|---|---|---|---|---|---|
| 1 | Claude Fable 5 Anthropic Best Overall | 59.9 | 95.0% | $10 / $50 | No change data |
| 2 | Claude Opus 4.8 Anthropic | 55.7 | 88.6% | $5 / $25 | No change data |
| 3 | Claude Sonnet 5 Anthropic Best Value | 53.4 | 85.2% | $3 / $15 | No change data |
| 4 | GPT-5.3 Codex OpenAI | — | 85.0% | — | No change data |
| 5 | Gemini 3.1 Pro Google | — | 80.6% | — | No change data |
| Rank | Model & Provider | Arena ELO | Cost / Image | Weekly Change |
|---|---|---|---|---|
| 1 | GPT Image 2 OpenAI Best Overall | 1385 | ~$0.05* | No change data |
| 2 | Reve 2.1 Reve | 1302 | — | No change data |
| 3 | Muse Image Meta | 1280 | — | No change data |
| 4 | Reve 2.0 Reve Best Value | 1271 | $0.024 | No change data |
| 5 | Gemini 3.1 Flash Image Google | 1261 | $0.067 | No change data |
| Rank | Model & Provider | Arena ELO | Cost / Second | Weekly Change |
|---|---|---|---|---|
| 1 | Gemini Omni Flash Google Best Overall | 1527 | $0.10 | No change data |
| 2 | Seedance 2.0 ByteDance | 1482 | $0.15 | No change data |
| 3 | Muse Video Meta | 1459 | — | No change data |
| 4 | HappyHorse 1.0 Alibaba | 1430 | $0.22 | No change data |
| 5 | Sora 2 Pro OpenAI | 1366 | $0.30 | No change data |
