Compare models
Put up to 4 models beside each other — token prices, context windows, capabilities and provider, from the same catalogue the model pages read.
| Attribute | Claude Fable 5.2claude-fable-5.2 | Kimi K2 Thinkingkimi-k2-thinking |
|---|---|---|
| Pricing | ||
| Input | $8.00 / 1M | $0.34 / 1M |
| Output | $40.00 / 1M | $1.50 / 1M |
| Cache Write (5m) | $10.00 / 1M | $0.34 / 1M |
| Cache Write (1h) | $16.00 / 1M | $0.34 / 1M |
| Cache Read | $0.20 / 1M | $0.34 / 1M |
| Web Search | $0 / 1M | $0 / 1M |
| Context | ||
| Max context | 1M | 262.1K |
| Max output | N/A | N/A |
| Capabilities | ||
| Vision | Yes | Yes |
| Function Calling | Yes | Yes |
| JSON Mode | Yes | Yes |
| Streaming | Yes | Yes |
| Catalogue | ||
| Provider | Anthropic | Moonshot AI |
| Category | chat | chat |
| Charge type | Pay As You Go | Pay As You Go |
| Released | — | — |
| Description | ||
| Summary | Claude Fable 5.2 is Anthropic's most capable model for advanced technical work and open-ended problem solving, delivering significant improvements in coding, reasoning, and vision over Fable 5.1. It excels at exceptionally complex tasks, including large-scale software migrations, electronic system design, and advanced scientific and mathematical research. With up to 30% faster output generation, clearer communication, and always-on adaptive extended thinking, Fable 5.2 supports configurable reasoning effort from low to max (default high), with thinking preserved across turns. Designed for frontier engineering, scientific discovery, and ambitious long-horizon agentic workflows, it complements Opus 5.5 as a specialized option for the most demanding technical challenges. | Kimi K2 Thinking is Moonshot AI's most advanced open reasoning model, built on a trillion-parameter MoE with 32B active parameters and a 256K context window. It's optimized for long-horizon, step-by-step reasoning with dynamic tool use, enabling sustained autonomous research, coding, and writing over hundreds of turns. It sets new open-source records on benchmarks like HLE and LiveCodeBench, and maintains stable multi-agent, tool-heavy workflows (200–300 calls) while balancing deep reasoning with efficient inference. |