Compare AI models with clarity
Hungry for Data, Open for All
Four boards—models, agents, LLMs, and toolchains—in one comparable frame. Rankings are generated via versioned pipelines with traceable sources and methodology.
⌘K / Ctrl+K or / to focus
No direct match found. Try model name, vendor, or tool keyword.
Try a shorter keyword, switch category terms, or search by vendor name.
Request this model?- DeepSeek V4.1 Flash: CED 552B, vendor agent scores, and the V4.1 Pro that is not listed
- Cursor’s own models: Auto as the floor, Composer / Grok, and when to pay for Opus 4.6
- Qwen3.8-27B: a 24GB consumer GPU against Opus 4.6’s coding table
- Build a local coding model: stack, weights, VRAM bands
- DeepSeek V4 Flash 0731 deep dive: 1M-context MoE at $0.09/$0.18 on OpenRouter
- Claude Opus 5 deep dive: Anthropic’s reasoning-and-coding flagship at roughly half of Fable
- Kimi K3 deep dive: how far the 2.8T open flagship sits from Claude and GPT
- How to choose AI infrastructure: a practical guide to model API hosts
- Qwen model series deep dive: a selection guide from Qwen2.5 to Qwen3.7
- Grok 4.5 deep dive: xAI’s flagship for coding and STEM
Global rankings
View all rankings| Rank | Name | Score | Key metric | 1M tokens (avg) |
|---|---|---|---|---|
| 1 | SpaceXAI: Grok 4.20 Multi-Agent | 99.9 | 2.0M ctx | $1.88 |
| 2 | SpaceXAI: Grok 4.20 | 99.9 | 2.0M ctx | $1.88 |
| 3 | Meta: Llama 4 Scout | 90.8 | 1.3M ctx | $0.20 |
| 4 | Thinking Machines: Inkling Small (free) | 86.9 | 1.0M ctx | $0.00 |
| 5 | Google: Lyria 3 Pro Preview | 86.9 | 1.0M ctx | $0.00 |
| 6 | DeepSeek: DeepSeek V4 Flash Latest | 86.9 | 1.0M ctx | $0.07 |
| 7 | OpenAI: GPT-6 Luna Pro (batch) | 86.8 | 1.1M ctx | $0.15 |
| 8 | Xiaomi: MiMo-V2.6-Flash | 86.8 | 1.1M ctx | $0.21 |
| 9 | Z.ai: GLM 5.3 Flash (batch) | 86.8 | 1.0M ctx | $0.13 |
| 10 | Z.ai: GLM Flash Latest | 86.8 | 1.0M ctx | $0.13 |
List prices in this table use OpenRouter as the primary listing source. Official vendor rows are on the Token hub. Fetched: 2026-09-30T11:29:46.224Z.
Core catalog refreshed:
Today's signals
Pick by task
-
Low-latency support
Bias toward speed and stability for high-volume support and FAQ automation.
Open speed preset -
Local coding model
Unlimited tokens. Absolute privacy. Built for heavy users—stack, weights, and VRAM before you buy the wrong GPU.
Open local coding hub -
Cost-sensitive batch
For offline generation and bulk rewrite; optimize for 1M-token cost.
Open cost preset