Claude Haiku 4.5
Anthropic
Snappy, inexpensive model for classification and routing.
- Context window
- 200K tokens
- Input / 1M tokens
- $0.80
- Output / 1M tokens
- $4.00
anthropic/claude-haiku-4.5Model catalog / Preview
Compare tasks, context windows and reference rates. This preview catalog is for exploration; models and commercial terms are confirmed individually for each pilot.
Reference rates, not a quote. Copy a model ID to use in the model field of an API request; copying does not enable access.
Anthropic
Snappy, inexpensive model for classification and routing.
anthropic/claude-haiku-4.5Anthropic
Deepest reasoning and long-horizon coding from Anthropic.
anthropic/claude-opus-4.8Anthropic
Balanced speed, cost, and quality for everyday agents.
anthropic/claude-sonnet-4.6Mistral
Code-specialized model tuned for fill-in-the-middle and repos.
mistral/codestral-2Cohere
Enterprise RAG-optimized model with tool use.
cohere/command-aDeepSeek
Open reasoning model with visible chain-of-thought.
deepseek/deepseek-r1DeepSeek
Cost-efficient reasoning model popular for coding.
deepseek/deepseek-v3.2OpenAI
High-quality text embeddings for search and RAG.
openai/text-embedding-3-largeBlack Forest
State-of-the-art open image model with precise prompt adherence.
bfl/flux-2-proUltra-cheap long context for retrieval-heavy pipelines.
google/gemini-2.5-flashMillion-token context with strong multimodal grounding.
google/gemini-2.5-proOpenAI
Fast, instruction-following image generation and editing.
openai/gpt-image-2OpenAI
Frontier reasoning and tool use across text and images.
openai/gpt-5OpenAI
Fast, low-cost model for high-volume production workloads.
openai/gpt-5-miniOpenAI
Smallest GPT-5 tier for classification, routing, and autocomplete.
openai/gpt-5-nanoxAI
Real-time-aware model with strong reasoning.
xai/grok-4xAI
Latency-optimized Grok tier for interactive products.
xai/grok-4-fastPhotorealistic text-to-image generation.
google/imagen-4Moonshot
Open trillion-parameter MoE with excellent tool use.
moonshot/kimi-k2Meta
Open-weight MoE model with strong general performance.
meta/llama-4-maverickMeta
10M-token context window for whole-corpus analysis.
meta/llama-4-scoutMistral
European frontier model with strong multilingual support.
mistral/mistral-large-3Qwen
Agentic coding model with strong repository-scale edits.
qwen/qwen3-coderQwen
Alibaba's flagship multilingual model.
qwen/qwen3-maxOpenAI
Text- and image-conditioned video generation with audio.
openai/sora-2Cinematic video generation with native soundtrack.
google/veo-3.1OpenAI
Robust multilingual speech-to-text transcription.
openai/whisper-3Need help choosing? Tell us about your workload.
Discuss your use case