AI Model Intelligence: Rankings, Benchmarks & Pricing
RuntimeWire's model intelligence desk: our own benchmark standings, head-to-head verdicts, and newsroom coverage — plus live pricing, context windows, and usage for every major AI model across text, image, video, audio, and speech.
Browse models by output type: text, image, video, audio, and speech. Each listing includes context window, supported input and output modalities, and current pricing — plus RuntimeWire's own signals: Composite Index rank, head-to-head record, and newsroom coverage.
RuntimeWire Composite Index — current standings
- Z.ai: GLM 5.3 Flash — 81.9 (2/3 suites)
- Claude Opus 5 (Fast) — 80.5 (2/3 suites)
- Qwen: Qwen3.8 Flash — 79.6 (2/3 suites)
- Google: Gemini 3.7 Flash — 78.1 (2/3 suites)
- step-3.7-flash — 75.3 (2/3 suites)
- DeepSeek-V4-Flash-0731 — 71.7 (1/3 suites)
- SpaceXAI: Grok 4.6 — 66.5 (1/3 suites)
- Anthropic: Claude Opus 4.8 — 62.0 (1/3 suites)
Latest head-to-head verdicts
- AuraFlow vs. Cosmos 3 Super — too close to call
- AnimateDiff Turbo vs. Wan 3.0 — Wan 3.0 wins (clear)
- Glm Image vs. ImagineArt 1.5 Pro Preview — ImagineArt 1.5 Pro Preview wins (clear)
- DeepSeek-V3.2 vs. Phi-4-mini-instruct — DeepSeek-V3.2 wins (clear)
- LTX Video (preview) vs. Seedance 2 Image to Video — Seedance 2 Image to Video wins (clear)
- grok-4.6 vs. Codestral-2501 — grok-4.6 wins (clear)
Most covered in RuntimeWire stories (last 30 days)
- OpenAI: GPT-5.4 — 73 stories; latest: Head to head: AnimateDiff Turbo vs Wan 3.0
- OpenAI: GPT-5.6 Sol — 39 stories; latest: OpenAI's safeguards missed 1,200 agents coordinating before 700 hacked Hugging Face
- Minimax — 26 stories; latest: Z.ai will release GLM-5.3 weights on August 28th after safety delay
- SpaceXAI: Grok 4.6 — 20 stories; latest: Head to head: grok-4.6 vs Codestral-2501
- MoonshotAI: Kimi K3 — 16 stories; latest: Moonshot seeks up to 30% of Kimi K3 sales on Azure, AWS and Google Cloud
- DeepSeek-V4-Flash — 15 stories; latest: Head to head: Google: Gemini 3.7 Flash vs DeepSeek-V4-Flash
Pricing and usage data is sourced from OpenRouter; benchmark standings, head-to-head verdicts, and story coverage are RuntimeWire's own reporting. Prices change frequently — confirm on the provider's site before relying on them.