Chat (LLM)
deepseek/deepseek-v4-flash- Input
- $0.15
- Output
- $0.31
- Cache read
- $0.003
deepseek/deepseek-v4-pro- Input
- $0.48
- Output
- $0.96
- Cache read
- $0.004
anthropic/claude-4-7-sonnet- Input
- $3.00
- Output
- $15.00
- Cache read
- $0.300
anthropic/claude-4-7-opus- Input
- $5.00
- Output
- $25.00
- Cache read
- $0.500
openai/gpt-5- Input
- $1.25
- Output
- $10.00
- Cache read
- $0.125
openai/gpt-5-mini- Input
- $0.25
- Output
- $2.00
- Cache read
- $0.025
google/gemini-3-pro- Input
- $1.25
- Output
- $10.00
openai/gpt-4.1-mini- Input
- $0.40
- Output
- $1.60
openai/gpt-4.1-nano- Input
- $0.10
- Output
- $0.40
openai/gpt-5-nano- Input
- $0.05
- Output
- $0.40
openai/gpt-5-codex- Input
- $1.25
- Output
- $10.00
openai/o4-mini- Input
- $1.10
- Output
- $4.40
google/gemini-3-flash- Input
- $0.30
- Output
- $2.50
google/gemini-3-flash-lite- Input
- $0.25
- Output
- $1.50
zai/glm-4-6- Input
- $0.06
- Output
- $0.22
- Cache read
- $0.011
alibaba/qwen3-coder-plus- Input
- $0.54
- Output
- $2.16
- Cache read
- $0.108
qwen/qwen3-vl-flash- Input
- $0.02
- Output
- $0.21
- Cache read
- $0.004
qwen/qwen3-vl-plus- Input
- $0.14
- Output
- $1.37
- Cache read
- $0.027
Embeddings
dashscope/text-embedding-v4Qwen3-Embedding family — Chinese-first dense text embeddings
Speech (TTS)
Transcribe (ASR)
Image generation
rx-image-z-proZ-image (DashScope direct) — reliable, low-latency premium tier
rx-image-qwen-editQwen-Image-Edit — reference-image conditioned generation (0–3 refs)
rx-image-gpt-draftGPT Image (draft) — cheapest tier, for composition search
rx-image-gptGPT Image — strong prompt adherence, text-to-image or 0–3 refs
rx-image-gpt-proGPT Image (high) — maximum fidelity, for final renders
Video generation
rx-video-ltxLTX-2.3 — fastest image-to-video, 720p-friendly
rx-video-wanWan2.2-Lightning — higher-quality image-to-video
rx-video-h3MiniMax H3 — text/reference to video with native stereo audio
rx-video-animateWan 2.2 Animate — copy a driving video's motion onto a character image
rx-avatar-talkInfiniteTalk — portrait + speech to a lip-synced talking video (length follows the audio)
rx-avatar-dubInfiniteTalk — re-voice an existing clip, keeping its framing and camera movement