GreenPT Docs

Model Cards

Full GreenPT model catalog (chat, coding, embeddings, speech-to-text, and reranking) with pricing and EU/US inference availability per model.

Token prices are in EUR per 1M tokens; speech-to-text models are billed per hour of audio. To list models programmatically, see the Models API.

Catalog

10 models · EU

api.greenpt.ai/v1 (EU default) · Region Inference

kimi-k3Moonshot AI

Moonshot Kimi K3, an open-weight 2.8T-parameter native multimodal agentic model with hybrid reasoning for long-horizon coding, knowledge work, and reasoning.

Input€3.30
Cached input€0.825
Output€16.50
Context1M
ReleasedJul 2026
ChatHybrid reasoningAgentic codingVision
glm-5.3z.ai

The newest z.ai flagship reasoning model, successor to glm-5.2, with agentic tool-use at the same price.

Input€1.10
Cached input€0.275
Output€4.40
Context1M
ReleasedAug 2026
ChatReasoningTool useAgentic coding
glm-5.3-flashz.ai

The light tier of GLM-5.3: the same 1M-token context window and reasoning ladder at a tenth of the price. Reasoning is on by default, so allow enough output budget for it.

Input€0.11
Cached input€0.022
Output€0.44
Context1M
ReleasedSep 2026
ChatReasoningTool useLong context
deepseek-v4.1-flashDeepSeek

DeepSeek V4.1 Flash, a multimodal reasoning model with a 1M-token context window for chat and agentic coding.

Input€0.22
Cached input€0.011
Output€1.10
Context1M
ReleasedSep 2026
ChatReasoningVisionLong context
minimax-m3MiniMax

MiniMax M3, a reasoning model with image input, tool calling and a 1M-token context window for agentic coding and long-horizon workflows.

Input€0.44
Cached input€0.11
Output€2.20
Context1M
Max output64k
ReleasedJul 2026
ChatReasoningAgentic codingVision
green-embeddingGreenPT
Backed by Qwen3-Embedding-4B

Dense multilingual embedding model for semantic search, retrieval, and RAG pipelines. Supports Matryoshka Representation Learning, so output dimensions are configurable from 32 up to 2560.

Input€0.20
OutputNo output charge
Context32k
Dimensionsup to 2560
ReleasedJun 2025
Dense embeddingsMultilingual (100+ languages)Semantic searchRAGMRL (custom dimensions)Instruction-aware
qwen3-embedding-8bQwen
Backed by Qwen3-Embedding-8B

State-of-the-art multilingual embedding model for semantic search, retrieval, and RAG pipelines. Supports Matryoshka Representation Learning, so output dimensions are configurable from 32 up to 4096.

Input€0.25
OutputNo output charge
Context32k
Dimensionsup to 4096
Released2026
Dense embeddingsMultilingual (100+ languages)Semantic searchRAGMRL (custom dimensions)Instruction-aware
green-sGreenPT

Speech-to-text model for pre-recorded and live transcription.

Input€0.23/hr
Output€0.31/hr live
Released2025
Pre-recorded transcriptionLive transcription
green-s-proGreenPTPro

Advanced speech-to-text model with multilingual transcription options.

Input€0.23/hr recorded
Output€0.31/hr live
Released2025
Pre-recorded transcriptionLive transcriptionMultilingual input €0.28/hrMultilingual live €0.37/hr
green-rerankGreenPT
Backed by Qwen3-Reranker-4B

Cross-encoder reranking model that scores and reorders search or RAG results by relevance to a query.

Input€0.12
OutputNo output charge
Context32k
ReleasedJun 2025
Document rerankingRAG result scoringMultilingual (100+ languages)Instruction-aware

Not recommended model ids (17)

These ids still answer, but a newer model does the job better. Open Details for an id's price and limits.

bge-multilingual-gemma2BAAINot recommended
use qwen3-embedding-8b
deepseek-v4-flash-0731DeepSeekNot recommended
use deepseek-v4.1-flash
gemma4GoogleNot recommended
glm-5.2z.aiNot recommended
use glm-5.3
gpt-oss-120bOpenAINot recommended
green-lGreenPTNot recommended
green-l-rawGreenPTNot recommended
green-rGreenPTNot recommended
green-r-rawGreenPTNot recommended
kimi-k2.6Moonshot AINot recommended
kimi-k2.7-codeMoonshot AINot recommended
llama-3.3-70b-instructMetaNot recommended
mistral-medium-3.5-128bMistralNot recommended
mistral-small-3.2-24b-instruct-2506MistralNot recommended
use glm-5.3-flash
qwen3-235b-a22b-instruct-2507QwenNot recommended
qwen3.5-397b-a17bQwenNot recommended
qwen3.6-35b-a3bQwenNot recommended

Deprecated model ids (10)

An id under a deprecation notice answers until the listed date. A deprecated id is no longer listed by GET /v1/models, requests to it return 400 Unsupported model, and a direct GET /v1/models/{id} lookup returns 404.

pixtral-12b-2409MistralDeprecated
end of life 24 September 2026 · use glm-5.3-flash
qwen3-coder-30b-a3b-instructQwenDeprecated
end of life 24 September 2026 · use deepseek-v4.1-flash
devstral-2-123b-instruct-2512MistralDeprecated
end of life 1 September 2026 · use glm-5.3
gemma-3-27b-itGoogleDeprecated
end of life 1 September 2026 · use deepseek-v4.1-flash
holo2-30b-a3bH CompanyDeprecated
end of life 1 September 2026 · use deepseek-v4.1-flash
voxtral-small-24b-2507MistralDeprecated
end of life 2 August 2026 · use green-s-pro
deepseek-r1-distill-llama-70bDeepSeekDeprecated
end of life 16 April 2026 · use deepseek-v4.1-flash
mistral-nemo-instruct-2407MistralDeprecated
end of life 16 April 2026 · use deepseek-v4.1-flash
devstral-small-2505MistralDeprecated
end of life 14 November 2025 · use deepseek-v4.1-flash
llama-3.1-8b-instructMetaDeprecated
end of life 14 November 2025 · use deepseek-v4.1-flash

Model changes

Every release, price change and status change, newest first.

  1. 1 October 2026
    green-sPrice

    The launch promo ended. Per hour of audio, pre-recorded from EUR 0.12 to 0.23, live from 0.16 to 0.31.

  2. 1 October 2026
    green-s-proPrice

    The launch promo ended. Per hour of audio, monolingual pre-recorded from EUR 0.12 to 0.23 and live from 0.16 to 0.31; multilingual pre-recorded from 0.14 to 0.28 and live from 0.19 to 0.37.

  3. 30 September 2026
    metisNew model
  4. 26 September 2026
    glm-5.2Price

    Moved to TensorX. Per 1M tokens, input from EUR 1.25 to 1.55, cached input from 0.30 to 0.40, output from 4.30 to 4.60.

  5. 24 September 2026
    bge-multilingual-gemma2Price

    Input from EUR 0.05 to 0.11 per 1M tokens.

  6. 24 September 2026
    glm-5.2Price

    Moved to Inceptron. Per 1M tokens, input from EUR 1.10 to 1.25, cached input from 0.275 to 0.30, output from 4.40 to 4.30.

  7. 24 September 2026
    kimi-k2.7-codePrice

    Cached input from EUR 0.165 to 0.198 per 1M tokens.

  8. 24 September 2026
    pixtral-12b-2409Deprecated

    End of life; requests return 400 Unsupported model. Use glm-5.3-flash instead.

  9. 24 September 2026
    qwen3-coder-30b-a3b-instructDeprecated

    End of life; requests return 400 Unsupported model. Use deepseek-v4.1-flash instead.

  10. 22 September 2026
    bge-multilingual-gemma2Not recommended

    Moved out of the recommended catalogue.

  11. 22 September 2026
    deepseek-v4-flash-0731Not recommended

    Moved out of the recommended catalogue.

  12. 22 September 2026
    gemma4Not recommended

    Moved out of the recommended catalogue.

  13. 22 September 2026
    glm-5.2Not recommended

    Moved out of the recommended catalogue.

  14. 22 September 2026
    gpt-oss-120bNot recommended

    Moved out of the recommended catalogue.

  15. 22 September 2026
    green-lNot recommended

    Moved out of the recommended catalogue.

  16. 22 September 2026
    green-l-rawNot recommended

    Moved out of the recommended catalogue.

  17. 22 September 2026
    green-rNot recommended

    Moved out of the recommended catalogue.

  18. 22 September 2026
    green-r-rawNot recommended

    Moved out of the recommended catalogue.

  19. 22 September 2026
    kimi-k2.6Not recommended

    Moved out of the recommended catalogue.

  20. 22 September 2026
    kimi-k2.7-codeNot recommended

    Moved out of the recommended catalogue.

  21. 22 September 2026
    llama-3.3-70b-instructNot recommended

    Moved out of the recommended catalogue.

  22. 22 September 2026
    mistral-medium-3.5-128bNot recommended

    Moved out of the recommended catalogue.

  23. 22 September 2026
    mistral-small-3.2-24b-instruct-2506Not recommended

    Moved out of the recommended catalogue.

  24. 22 September 2026
    qwen3-235b-a22b-instruct-2507Not recommended

    Moved out of the recommended catalogue.

  25. 22 September 2026
    qwen3.5-397b-a17bNot recommended

    Moved out of the recommended catalogue.

  26. 22 September 2026
    qwen3.6-35b-a3bNot recommended

    Moved out of the recommended catalogue.

  27. 11 September 2026
    deepseek-v4.1-flashNew model
  28. 1 September 2026
    devstral-2-123b-instruct-2512Deprecated

    End of life; requests return 400 Unsupported model. Use glm-5.3 instead.

  29. 1 September 2026
    gemma-3-27b-itDeprecated

    End of life; requests return 400 Unsupported model.

  30. 1 September 2026
    holo2-30b-a3bDeprecated

    End of life; requests return 400 Unsupported model.

  31. 31 August 2026
    glm-5.3New model
  32. 31 August 2026
    glm-5.3-flashNew model
  33. 3 August 2026
    deepseek-v4-flash-0731New model
  34. 2 August 2026
    voxtral-small-24b-2507Deprecated

    End of life; requests return 400 Unsupported model. Use green-s-pro instead.

  35. 27 July 2026
    kimi-k3New model
  36. 7 July 2026
    minimax-m3New model
  37. 30 June 2026
    kimi-k2.6New model
  38. 30 June 2026
    kimi-k2.7-codeNew model
  39. 29 June 2026
    devstral-2-123b-instruct-2512New model
  40. 29 June 2026
    glm-5.2New model
  41. 26 June 2026
    qwen3-embedding-8bNew model
  42. 9 June 2026
    holo2-30b-a3bNew model
  43. 9 June 2026
    mistral-medium-3.5-128bNew model
  44. 9 June 2026
    qwen3.6-35b-a3bNew model
  45. 20 April 2026
    gemma-3-27b-itNew model
  46. 20 April 2026
    gemma4New model
  47. 16 April 2026
    deepseek-r1-distill-llama-70bDeprecated

    End of life; requests return 400 Unsupported model.

  48. 16 April 2026
    mistral-nemo-instruct-2407Deprecated

    End of life; requests return 400 Unsupported model.

  49. 1 April 2026
    bge-multilingual-gemma2New model
  50. 1 April 2026
    pixtral-12b-2409New model
  51. 18 March 2026
    qwen3.5-397b-a17bNew model
  52. 14 November 2025
    devstral-small-2505Deprecated

    End of life; requests return 400 Unsupported model.

  53. 14 November 2025
    llama-3.1-8b-instructDeprecated

    End of life; requests return 400 Unsupported model.

  54. 7 October 2025
    deepseek-r1-distill-llama-70bNew model
  55. 7 October 2025
    devstral-small-2505New model
  56. 7 October 2025
    gpt-oss-120bNew model
  57. 7 October 2025
    llama-3.1-8b-instructNew model
  58. 7 October 2025
    llama-3.3-70b-instructNew model
  59. 7 October 2025
    mistral-nemo-instruct-2407New model
  60. 7 October 2025
    mistral-small-3.2-24b-instruct-2506New model
  61. 7 October 2025
    qwen3-235b-a22b-instruct-2507New model
  62. 7 October 2025
    qwen3-coder-30b-a3b-instructNew model
  63. 7 October 2025
    voxtral-small-24b-2507New model
  64. 10 September 2025
    green-embeddingNew model
  65. 10 September 2025
    green-rerankNew model
  66. 10 September 2025
    green-s-proNew model
  67. 9 June 2025
    green-sNew model
  68. 9 May 2025
    green-lNew model
  69. 9 May 2025
    green-l-rawNew model
  70. 9 May 2025
    green-rNew model
  71. 9 May 2025
    green-r-rawNew model

Built-in system prompts

Some ids carry a system prompt of their own, so the id you send decides which rules the model gets on top of yours:

  • Compression models (glm-5.2-caveman, -honey and -ponytail, each also in -lite and -ultra) carry an output-compression ruleset: the same model and price per token as glm-5.2, with a shorter answer. See Compression models.
  • GreenPT models (green-l, green-r) carry GreenPT's Sustainability system prompt, optimized for our use cases; their raw twins (green-l-raw, green-r-raw) carry none, so yours is the only system prompt sent to the model.
  • Every other id in the catalog carries no system prompt.

When a model carries one and you send one too, both go to the model as a single system message: yours first, the model's rules last. Picking the id is an explicit opt-in to that style, so on a conflict the model's rules win over a generic prompt your client happens to send. A developer message counts as a system message here and merges the same way.

Catalog Notes

All models are served through the GreenPT API proxy with an OpenAI-compatible interface. Prices are in EUR per million tokens unless noted otherwise, and API pricing applies to API usage only. For how we measure the environmental footprint of a request, see Carbon calculations.

Models API

GreenPT Models API. List the models available through the OpenAI-compatible endpoint and their capabilities.

Chat Completion

Create a chat completion with a GreenPT model. We recommend glm-5.3.

On this page

CatalogBuilt-in system promptsCatalog Notes