MI modeļu līderu tabula
Every model you can run on Railwail right now, by category: quality from the public Arena leaderboard, the live price of a default call, the context window and run times measured on our own jobs. Scores only where the arena rates exactly the model we run.
Arena scores published 2026. gada 22. sept.Run times from Railwail jobs, last 90 days
- Models you can run
- 274
- Labs
- 34
- Categories
- 9
Today’s record holders
Best models by category
Quality from the public Arena, prices from our pricing rules. Open a category to sort by quality, price or value.
Text & chat
16 of 30 rated- Best quality
- Gemini 3.1 ProArena 1487
- Cheapest
- Granite Code 8B0,12 $ / 1M tokens
- Best value
- Gemini 3.1 Pro8 pts/$
Image generation
22 of 66 rated- Best quality
- Qwen Image 3 ProArena 1255
- Cheapest
- Stable Diffusion XL≈ 0,0015 $/palaidi
- Best value
- Nano Banana 2 Lite3828 pts/$
Image editing
7 of 7 rated- Best quality
- Grok Imagine ImageArena 1329
- Cheapest
- Grok Imagine Image0,024 $/attēls
- Best value
- Grok Imagine Image3679 pts/$
Video generation
13 of 27 rated- Best quality
- Google Veo 3.1Arena 1362
- Cheapest
- LTX-Video (Lightricks)≈ 0,0229 $/palaidi
- Best value
- Grok Imagine Video504 pts/$
Image to video
14 of 14 rated- Best quality
- Grok Imagine VideoArena 1415
- Cheapest
- Seedance Lite0,0432 $/s
- Best value
- Grok Imagine Video556 pts/$
9 models
Filter by labTap a lab to filter every board.
01
Language models
30Chat, reasoning and code models billed per token. Sorted by Arena score, highest first; models without a rating follow by price.
| Actions | ||||||
|---|---|---|---|---|---|---|
Claude Opus 4.7 Arena 1494 · 30,00 $/1M izvade · 1 milj. ctx | 1494 | 6,00 $/1M ievade6,00 $ | 30,00 $ | 1 milj. | · | Try |
Gemini 3.1 Pro Arena 1487 · 14,40 $/1M izvade · 1 milj. ctx | 1487 | 2,40 $/1M ievade2,40 $ | 14,40 $ | 1 milj. | · | Try |
GPT-5.5 Arena 1476 · 36,00 $/1M izvade · 400 tūkst. ctx | 1476 | 6,00 $/1M ievade6,00 $ | 36,00 $ | 400 tūkst. | · | Try |
Gemini 3 Flash Arena 1474 · 3,60 $/1M izvade · 1 milj. ctx | 1474 | 0,60 $/1M ievade0,60 $ | 3,60 $ | 1 milj. | · | Try |
Claude Opus 4.8 Arena 1473 · 30,00 $/1M izvade · 1 milj. ctx · median 2,3 s | 1473 | 6,00 $/1M ievade6,00 $ | 30,00 $ | 1 milj. | 2,3 s15 runs | Try |
Claude Sonnet 4.6 Arena 1473 · 18,00 $/1M izvade · 1 milj. ctx | 1473 | 3,60 $/1M ievade3,60 $ | 18,00 $ | 1 milj. | · | Try |
GPT-5.4 Arena 1466 · 18,00 $/1M izvade · 1,1 milj. ctx | 1466 | 3,00 $/1M ievade3,00 $ | 18,00 $ | 1,1 milj. | · | Try |
Gemini 2.5 Pro Arena 1446 · 12,00 $/1M izvade · 1 milj. ctx | 1446 | 1,50 $/1M ievade1,50 $ | 12,00 $ | 1 milj. | · | Try |
GPT-5.1 Arena 1439 · 12,00 $/1M izvade · 400 tūkst. ctx | 1439 | 1,50 $/1M ievade1,50 $ | 12,00 $ | 400 tūkst. | · | Try |
OpenAI o3 Arena 1432 · 9,60 $/1M izvade · 200 tūkst. ctx | 1432 | 2,40 $/1M ievade2,40 $ | 9,60 $ | 200 tūkst. | · | Try |
| Claude Haiku 4.5 Anthropic | 1,20 $/1M ievade | |||||
| GPT-4.1 OpenAI | 2,40 $/1M ievade | |||||
| OpenAI o4-mini OpenAI | 1,32 $/1M ievade | |||||
| o3-mini OpenAI | 1,32 $/1M ievade | |||||
| GPT-4o OpenAI | 3,00 $/1M ievade | |||||
| GPT-4o Mini OpenAI | 0,18 $/1M ievade | |||||
| Granite Code 8B IBM | 0,060 $/1M ievade | |||||
| GPT-6 Luna OpenAI | 0,12 $/1M ievade | |||||
| Granite Code 20B IBM | 0,12 $/1M ievade | |||||
| GPT-5.4 Nano OpenAI | 0,24 $/1M ievade | |||||
| GPT-5 Mini OpenAI | 0,30 $/1M ievade | |||||
| DeepSeek V4 Flash DeepSeek | 0,36 $/1M ievade | |||||
| DeepSeek V4.1 Flash DeepSeek | 0,36 $/1M ievade | |||||
| GPT-5.4 Mini OpenAI | 0,90 $/1M ievade | |||||
| DeepSeek V4 Pro DeepSeek | 1,584 $/1M ievade | |||||
| Claude Sonnet 5 Anthropic | 2,40 $/1M ievade | |||||
| GPT-6 Sol OpenAI | 2,40 $/1M ievade | |||||
| Claude Opus 5.5 Anthropic | 4,80 $/1M ievade | |||||
| Claude Fable 5.1 Anthropic | 12,00 $/1M ievade | |||||
| GPT-6 Astra OpenAI | 12,00 $/1M ievade |
10 of 30 · Median run “·” = Fewer than 5 completed runs in 90 days
Arena scores: Text Arena leaderboard by LMArena, published 2026. gada 13. sept., CC BY 4.0.Ranked by quality, price and value →
A usage column appears once 5 models have runs from 3 or more different accounts in 30 days.
02
Image generation & editing
83/134Text-to-image, editing, upscaling and background removal. Sorted by Arena score, highest first; models without a rating follow by price.
| Actions | |||
|---|---|---|---|
Qwen Image 3 Pro Arena 1255 | 1255 | 0,048 $/attēls | Try |
Nano Banana 2 Lite Arena 1250 | 1250 | 0,0408 $/attēls | Try |
Ideogram V4 Quality Arena 1204 | 1204 | 0,12 $/attēls | Try |
Grok Imagine Image Arena 1171 | 1171 | 0,024 $/attēls | Try |
Recraft V4.1 Utility Pro Arena 1169 | 1169 | 0,30 $/attēls | Try |
Hunyuan Image 3 Arena 1151 | 1151 | 0,096 $/attēls | Try |
Nano Banana Arena 1150 | 1150 | 0,0468 $/attēls | Try |
Recraft V4.1 Pro Arena 1131 | 1131 | 0,30 $/attēls | Try |
Google Imagen 4 Arena 1129 | 1129 | 0,048 $/attēls | Try |
Qwen Image 2512 Arena 1125 | 1125 | 0,024 $/attēls | Try |
| Recraft V4 Recraft | 0,048 $/attēls | ||
| Flux Kontext Max Black Forest Labs | 0,096 $/attēls | ||
| Flux Kontext Pro Black Forest Labs | 0,048 $/attēls | ||
| Qwen Image Alibaba (Qwen) | 0,030 $/attēls | ||
| Ideogram v3 Quality Ideogram | 0,108 $/attēls | ||
| Photon Luma AI | 0,036 $/attēls | ||
| P Image Pruna AI | 0,0060 $/attēls | ||
| Gen4 Image Runway | 0,096 $/attēls | ||
| Recraft V3 Recraft | 0,048 $/attēls | ||
| FLUX 1.1 Pro Black Forest Labs | 0,048 $/attēls | ||
| Ideogram v2 Ideogram | 0,096 $/attēls | ||
| Stable Diffusion 3.5 Large Stability AI | 0,078 $/attēls | ||
| 851-Labs Background Remover Community | ≈ 0,00050 $/palaidi | ||
| Remove Background (lucataco) Community | ≈ 0,00060 $/palaidi | ||
| Remove Object (LaMa) Community | ≈ 0,00090 $/palaidi | ||
| DDColor Community | ≈ 0,0012 $/palaidi | ||
| Html To Image Community | 0,0012 $/attēls | ||
| Stable Diffusion XL Stability AI | ≈ 0,0015 $/palaidi | ||
| Vectorizer (VTracer) Community | ≈ 0,0022 $/palaidi | ||
| Real-ESRGAN 4x Community | 0,0024 $/attēls | ||
| SDXL Inpainting Community | ≈ 0,0026 $/palaidi | ||
| GFPGAN v1.4 Tencent ARC | ≈ 0,0027 $/palaidi | ||
| BiRefNet Background Removal Community | ≈ 0,0036 $/palaidi | ||
| FLUX.1 [schnell] Black Forest Labs | 0,0036 $/attēls | ||
| FLUX.1 Redux Black Forest Labs | 0,0036 $/attēls | ||
| Consistent Character Community | ≈ 0,0038 $/palaidi | ||
| InstructPix2Pix Community | ≈ 0,0039 $/palaidi | ||
| CodeFormer Community | ≈ 0,0041 $/palaidi | ||
| Rembg Community | ≈ 0,0047 $/palaidi | ||
| Real-ESRGAN Anime 4x Community | ≈ 0,0051 $/palaidi | ||
| Sticker Maker Community | ≈ 0,0055 $/palaidi | ||
| Flux Fast Community | 0,0060 $/attēls | ||
| Hidream L1 Fast Community | 0,0060 $/attēls | ||
| PhotoMaker Tencent ARC | ≈ 0,0080 $/palaidi | ||
| Swin2SR Community | ≈ 0,0088 $/palaidi | ||
| Icons (SDXL Flat Pop) Community | ≈ 0,0094 $/palaidi | ||
| Bringing Old Photos Back to Life Microsoft | ≈ 0,010 $/palaidi | ||
| Face to Many Community | ≈ 0,0105 $/palaidi | ||
| Image 01 MiniMax | 0,012 $/attēls | ||
| Photon Flash Luma AI | 0,012 $/attēls | ||
| Recraft Vectorize Recraft | 0,012 $/attēls | ||
| ControlNet Depth Community | ≈ 0,0144 $/palaidi | ||
| SDXL Emoji Community | ≈ 0,0156 $/palaidi | ||
| Janus Pro 7B DeepSeek | ≈ 0,0169 $/palaidi | ||
| FLUX PuLID Community | ≈ 0,018 $/palaidi | ||
| Lucid Origin Community | 0,0201 $/attēls | ||
| BRIA Remove Background Bria AI | 0,0216 $/attēls | ||
| Clarity Upscaler Community | ≈ 0,024 $/palaidi | ||
| Hunyuan Image 2.1 Tencent | 0,024 $/attēls | ||
| Wan 2.2 Image Community | 0,024 $/attēls | ||
| Recraft 20B Recraft | 0,0264 $/attēls | ||
| IDM-VTON (Virtual Try-On) Community | ≈ 0,0277 $/palaidi | ||
| IP-Adapter FaceID Plus v2 Community | ≈ 0,0277 $/palaidi | ||
| Face to Sticker Community | ≈ 0,0289 $/palaidi | ||
| Flux Krea Dev Black Forest Labs | 0,030 $/attēls | ||
| FLUX.1 [dev] Black Forest Labs | 0,030 $/attēls | ||
| FLUX.1 Canny [dev] Black Forest Labs | 0,030 $/attēls | ||
| FLUX.1 Depth [dev] Black Forest Labs | 0,030 $/attēls | ||
| FLUX.1 Kontext [dev] Black Forest Labs | 0,030 $/attēls | ||
| Ideogram V2A Turbo Ideogram | 0,030 $/attēls | ||
| Ideogram v3 Turbo Ideogram | 0,036 $/attēls | ||
| Qwen Image 3 Alibaba (Qwen) | 0,036 $/attēls | ||
| Qwen-Image-Edit Alibaba (Qwen) | 0,036 $/attēls | ||
| Seedream 4 ByteDance | 0,036 $/attēls | ||
| Wan 2.7 Image Alibaba (Wan) | 0,036 $/attēls | ||
| Wan 2.7 Image Pro Alibaba (Wan) | 0,036 $/attēls | ||
| LogoAI (SDXL Logo Generator) Community | ≈ 0,0372 $/palaidi | ||
| Flux Dev Black Forest Labs | 0,0384 $/attēls | ||
| Qwen Image 2 Alibaba (Qwen) | 0,042 $/attēls | ||
| Seedream 5 Lite ByteDance | 0,042 $/attēls | ||
| Stable Diffusion 3 Stability AI | 0,042 $/attēls | ||
| Stable Diffusion 3.5 Medium Stability AI | 0,042 $/attēls | ||
| TRELLIS (3D) Community | ≈ 0,042 $/palaidi | ||
| ESRGAN Classic Community | ≈ 0,0468 $/palaidi | ||
| Fibo Bria AI | 0,048 $/attēls | ||
| FLUX.1 Fill [dev] Black Forest Labs | 0,048 $/attēls | ||
| Grok Imagine Image 2 xAI | 0,048 $/attēls | ||
| Ideogram V2A Ideogram | 0,048 $/attēls | ||
| Image 3.2 Bria AI | 0,048 $/attēls | ||
| Professional Headshot (FLUX Kontext) Black Forest Labs | 0,048 $/attēls | ||
| Recraft V4.1 Recraft | 0,048 $/attēls | ||
| Recraft V4.1 Utility Recraft | 0,048 $/attēls | ||
| Seedream 4.5 ByteDance | 0,048 $/attēls | ||
| Stable Diffusion 3.5 Large Turbo Stability AI | 0,048 $/attēls | ||
| Recraft 20B SVG Recraft | 0,0528 $/attēls | ||
| Magnific-Style Upscaler Community | ≈ 0,0589 $/palaidi | ||
| FLUX.1 Canny Black Forest Labs | 0,060 $/attēls | ||
| FLUX.1 Depth Black Forest Labs | 0,060 $/attēls | ||
| FLUX.1 Fill Black Forest Labs | 0,060 $/attēls | ||
| Ideogram V2 Turbo Ideogram | 0,060 $/attēls | ||
| AuraFlow v0.3 Community | ≈ 0,0648 $/palaidi | ||
| Flux Pro Black Forest Labs | 0,066 $/attēls | ||
| Ad Inpaint (Product Photo) Community | ≈ 0,0672 $/palaidi | ||
| ControlNet Canny Community | ≈ 0,0697 $/palaidi | ||
| Point-E Community | ≈ 0,0708 $/palaidi | ||
| Flux 1.1 Pro Ultra Black Forest Labs | 0,072 $/attēls | ||
| Ideogram V3 Balanced Ideogram | 0,072 $/attēls | ||
| Ideogram V4 Balanced Ideogram | 0,072 $/attēls | ||
| Playground v2.5 (1024px Aesthetic) Playground AI | ≈ 0,0732 $/palaidi | ||
| IC-Light (Product Relighting) Community | ≈ 0,0781 $/palaidi | ||
| Nano Banana 2 Google DeepMind | 0,0804 $/attēls | ||
| InstantID Community | ≈ 0,084 $/palaidi | ||
| Rd Animation Community | 0,084 $/attēls | ||
| Kuaishou Kolors Community | ≈ 0,0889 $/palaidi | ||
| FLUX.1-dev Inpainting Community | ≈ 0,090 $/palaidi | ||
| Qwen Image 2 Pro Alibaba (Qwen) | 0,090 $/attēls | ||
| Cartoonify Community | ≈ 0,096 $/palaidi | ||
| Recraft v3 SVG Recraft | 0,096 $/attēls | ||
| Recraft V4 SVG Recraft | 0,096 $/attēls | ||
| Shap-E (OpenAI) Community | ≈ 0,1116 $/palaidi | ||
| Gpt Image 2 OpenAI | 0,1536 $/attēls | ||
| Hunyuan3D 2.0 Tencent | ≈ 0,156 $/palaidi | ||
| Gpt Image 1.5 OpenAI | 0,1632 $/attēls | ||
| Nano Banana Pro Google DeepMind | 0,18 $/attēls | ||
| OOTDiffusion (Try-On) Community | ≈ 0,1801 $/palaidi | ||
| Hunyuan3D 2.1 Community | ≈ 0,264 $/palaidi | ||
| CCSR (Content-Consistent SR) Community | ≈ 0,2761 $/palaidi | ||
| Gpt Image 2.5 Flare OpenAI | 0,30 $/attēls | ||
| Gpt Image 2.5 Sunburst OpenAI | 0,30 $/attēls | ||
| Recraft V4 Pro Recraft | 0,30 $/attēls | ||
| SUPIR Upscaler Community | ≈ 0,30 $/palaidi | ||
| SUPIR Community | ≈ 0,4921 $/palaidi | ||
| StarVector 8B (image-to-SVG) Community | 0,00117 $/GPU s | ||
| DreamGaussian Community | 0,00168 $/GPU s |
10 of 83
Arena scores: Text-to-Image Arena leaderboard by LMArena, published 2026. gada 22. sept., CC BY 4.0.Ranked by quality, price and value →
03
Video generation
31/46Text-to-video and image-to-video. Sorted by Arena score, highest first; models without a rating follow by price.
| Actions | ||||
|---|---|---|---|---|
Google Veo 3.1 Arena 1362 | 1362 | 0,48 $/s1,92 $ for 4 s | · | Try |
Google Veo 3.1 Fast Arena 1358 | 1358 | 0,18 $/s0,72 $ for 4 s | · | Try |
Grok Imagine Video Arena 1342 · median 48 s | 1342 | 0,060 $/s0,30 $ for 5 s | 48 s31 runs | Try |
PixVerse v5.6 Arena 1240 | 1240 | 0,084 $/s0,42 $ for 5 s | · | Try |
Runway Gen 4.5 Arena 1225 | 1225 | 0,144 $/s0,72 $ for 5 s | · | Try |
Hailuo 2.3 Arena 1206 | 1206 | 0,336 $/video | · | Try |
Seedance Pro Arena 1191 | 1191 | 0,18 $/s0,90 $ for 5 s | · | Try |
MiniMax Hailuo 02 Arena 1181 | 1181 | 0,324 $/video | · | Try |
Google Veo 2 Arena 1164 | 1164 | 0,60 $/s4,80 $ for 8 s | · | Try |
Kling v2.1 Master Arena 1163 | 1163 | 0,336 $/s1,68 $ for 5 s | · | Try |
| Seedance Lite ByteDance | 0,0432 $/s | |||
| Luma Ray-2 720p Luma AI | 0,216 $/s | |||
| Mochi 1 Genmo | ≈ 0,5041 $/palaidi | |||
| FILM Frame Interpolation Google Research | ≈ 0,0016 $/palaidi | |||
| Wav2Lip Community | ≈ 0,0067 $/palaidi | |||
| LTX-Video (Lightricks) Lightricks | ≈ 0,0229 $/palaidi | |||
| SwinIR Video Community | ≈ 0,0277 $/palaidi | |||
| Wan 2.2 5B Fast Alibaba (Wan) | 0,030 $/video | |||
| RIFE Frame Interpolation Community | ≈ 0,0444 $/palaidi | |||
| Wan 2.2 Image-to-Video Alibaba (Wan) | 0,060 $/video | |||
| Wan 2.2 Text-to-Video Alibaba (Wan) | 0,060 $/video | |||
| MuseTalk Community | ≈ 0,0624 $/palaidi | |||
| ToonCrafter Community | ≈ 0,0852 $/palaidi | |||
| LivePortrait Community | ≈ 0,0949 $/palaidi | |||
| SadTalker Community | ≈ 0,1165 $/palaidi | |||
| AnimateDiff Community | ≈ 0,1201 $/palaidi | |||
| VideoCrafter Community | ≈ 0,1561 $/palaidi | |||
| Kling v2.1 Kuaishou (Kling) | 0,060 $/s | |||
| Runway Gen-4 Turbo Runway | 0,060 $/s | |||
| Luma Ray Flash 2 Luma AI | 0,072 $/s | |||
| Seedance 1 Pro Fast ByteDance | 0,072 $/s | |||
| CogVideoX-5B (open) Zhipu AI | ≈ 0,4081 $/palaidi | |||
| MagicAnimate Community | ≈ 0,4081 $/palaidi | |||
| Kling V2.5 Turbo Pro Kuaishou (Kling) | 0,084 $/s | |||
| EchoMimic Community | ≈ 0,48 $/palaidi | |||
| Mochi 1 Community | ≈ 0,5041 $/palaidi | |||
| CogVideoX-5B Community | ≈ 0,60 $/palaidi | |||
| Minimax Video MiniMax | 0,60 $/video | |||
| Wan 3 Alibaba (Qwen) | 0,12 $/s | |||
| Google Veo 3 Fast Google DeepMind | 0,18 $/s | |||
| Kling v3 Kuaishou (Kling) | 0,2688 $/s | |||
| Kling v3 Omni Kuaishou (Kling) | 0,2688 $/s | |||
| HunyuanVideo Tencent | ≈ 3,06 $/palaidi | |||
| Google Veo 3 (Replicate) Google DeepMind | 0,48 $/s | |||
| AnimateDiff Lightning Community | 0,00117 $/GPU s | |||
| V-Express Community | 0,00168 $/GPU s |
10 of 31 · Median run “·” = Fewer than 5 completed runs in 90 days
Arena scores: Text-to-Video Arena leaderboard by LMArena, published 2026. gada 22. sept., CC BY 4.0.Ranked by quality, price and value →
04
Text to speech
7/14Voices from text, including voice cloning. Sorted by the price of one default call, lowest first.
| Actions | ||
|---|---|---|
| 0,018 $/1000 rakstzīmes | Try | |
| 0,024 $/1000 rakstzīmes | Try | |
| 0,030 $/1000 rakstzīmes | Try | |
| 0,036 $/1000 rakstzīmes | Try | |
| Kokoro TTS 82M Community | ≈ 0,00030 $/palaidi | |
| StyleTTS 2 Community | ≈ 0,00040 $/palaidi | |
| Parler-TTS Community | ≈ 0,0024 $/palaidi | |
| Spark TTS Community | ≈ 0,0035 $/palaidi | |
| ≈ 0,0157 $/palaidi | Try | |
| RVC Voice Conversion Community | ≈ 0,0505 $/palaidi | |
| ≈ 0,0576 $/palaidi | Try | |
| OpenVoice v2 Community | ≈ 0,0673 $/palaidi | |
| Tortoise TTS Community | ≈ 0,0816 $/palaidi | |
| ≈ 0,0972 $/palaidi | Try |
05
Speech to text
5Transcription and speaker diarization. Sorted by the price of one default call, lowest first.
06
Vision & multimodal
30Open models that caption, read or segment images, billed by GPU time. Sorted by the price of one default call, lowest first.
| Actions | ||
|---|---|---|
| ≈ 0,00030 $/palaidi | Try | |
| ≈ 0,00030 $/palaidi | Try | |
| ≈ 0,00030 $/palaidi | Try | |
| ≈ 0,00060 $/palaidi | Try | |
| ≈ 0,0012 $/palaidi | Try | |
| ≈ 0,0012 $/palaidi | Try | |
| ≈ 0,0012 $/palaidi | Try | |
| ≈ 0,0012 $/palaidi | Try | |
| ≈ 0,0014 $/palaidi | Try | |
| ≈ 0,0015 $/palaidi | Try | |
| MiniCPM-V 2.6 Community | ≈ 0,0015 $/palaidi | |
| Moondream2 Community | ≈ 0,0020 $/palaidi | |
| Qwen2-VL 7B Instruct Community | ≈ 0,0024 $/palaidi | |
| Llama 3.2 Vision 11B (Ollama) Community | ≈ 0,0039 $/palaidi | |
| Depth Anything v2 Community | ≈ 0,0050 $/palaidi | |
| Llama 3.2 Vision 90B Community | ≈ 0,0070 $/palaidi | |
| DeepSeek-VL 7B DeepSeek | ≈ 0,0086 $/palaidi | |
| GLPN Depth Community | ≈ 0,0094 $/palaidi | |
| CogVLM2 19B Community | ≈ 0,0114 $/palaidi | |
| Dots OCR Community | ≈ 0,0132 $/palaidi | |
| SAM 2 (Segment Anything 2) Meta | ≈ 0,018 $/palaidi | |
| olmOCR Community | ≈ 0,0205 $/palaidi | |
| PaddleOCR v3 Community | ≈ 0,0216 $/palaidi | |
| Mask2Former Meta | ≈ 0,0276 $/palaidi | |
| Lotus-G Community | ≈ 0,0408 $/palaidi | |
| CLIP Interrogator Community | ≈ 0,0457 $/palaidi | |
| Molmo 7B Community | ≈ 0,0505 $/palaidi | |
| Donut Document Community | ≈ 0,0648 $/palaidi | |
| Marigold Community | ≈ 0,0936 $/palaidi | |
| LLaVA 1.6 Vicuna 13B Community | ≈ 0,1021 $/palaidi |
10 of 30
07
Open-weight text & code
10Open models on dedicated GPUs, billed by GPU time. Sorted by the price of one default call, lowest first.
08
Music & audio
3Music and sound generation. Sorted by the price of one default call, lowest first.
09
Embeddings
2Vectors for search and retrieval, billed per token. Sorted by input price per 1M tokens, lowest first.
| Actions | |||
|---|---|---|---|
OpenAI text-embedding-3-small 8,2 tūkst. ctx | 0,024 $/1M ievade | 8,2 tūkst. | Try |
OpenAI text-embedding-3-large 8,2 tūkst. ctx | 0,156 $/1M ievade | 8,2 tūkst. | Try |
10
How this leaderboard works
Every figure comes from our catalog, our pricing rules, our own job records or, for quality, the public Arena leaderboard. What we cannot source, we leave out.
- What is listed
- Only models you can run right now: active and with a verified price. Left out: 130 catalog entries without a price (they cannot run), 0 removed models and 5 duplicate entries of the same provider model. Legacy models stay listed and link their successor.
- Price
- What one call with the model’s default settings costs on Railwail, from the same pricing rules that bill your runs (1 credit = $0.01). Token models show the price per 1M input and output tokens. Prices marked ≈ are billed by GPU time: the figure is a typical run, the charge follows the real run time.
- Median run time
- Measured on completed Railwail jobs: the median of the newest up to 500 runs in the last 90 days, shown from 5 runs. It includes queueing at the provider. For language models it depends on the answer length, so the LLM board is never sorted by speed. Measured today: 3 models.
- Quality
- Arena scores from the public Arena leaderboard (formerly LMArena, CC BY 4.0), where people compare two anonymous models on the same prompt and vote. A model gets a score only when the arena entry is exactly the model and default setting we run (reviewed list, 56 models today). No score is estimated.
- Context window
- The maximum tokens per request as listed in our catalog. Not re-measured by us.
- Usage
- Runs per model are only shown once they say something: at least 5 models with runs from 3 or more different accounts in 30 days. Until then there is no popularity ranking.
- What we don’t show
- No votes of our own and no estimated scores: our own arena has 0 decided battles and 2 model votes, far too few for a rating. Models the public Arena has not rated in the setting we run show “No Arena rating”.
Source:arena.ai leaderboardDatasetCC BY 4.0
Found a wrong price or a missing model? Write to us
Questions about the leaderboard
Which language model is the cheapest on Railwail?
Right now Granite Code 8B by IBM at 0,060 $/1M ievade (output 0,30 $/1M izvade). The language model board lists all 30 language models by input price.
Which model has the largest context window?
GPT-5.4 by OpenAI with 1,1 milj. tokens per request, as listed in our catalog.
What is the lowest price per image?
FLUX.1 [schnell] by Black Forest Labs at 0,0036 $/attēls. Models billed by GPU time can cost less per run; their price, marked ≈, is a typical run and the charge follows the real run time.
Where do the quality scores come from?
From the public Arena leaderboard (formerly LMArena): people compare two anonymous models on the same prompt and vote for the better result; the votes give each model a score. We use the published leaderboards (text: 2026. gada 13. sept., image and video: 2026. gada 22. sept.) under CC BY 4.0 and show a score only where the arena rates exactly the model and default setting that runs on Railwail.
How current are the prices?
They come from the same pricing rules that bill your runs and were last checked on 2026. gada 24. sept.. The snapshot on this page is refreshed about once an hour.
How do I run one of these models?
Open the model and try it on its page, or create an account, get an API key and call it through the Railwail API at https://railwail.com/api/v1. Your balance is prepaid in USD; runs are billed by the same pricing rules as the prices shown here.
Keep exploring
One key, one balance in USD, every model on this page.
Get an API key