Baltor Get started

Models

Inkling-Small

By thinkingmachines. 266.0 billion parameters, a context window of Unknown tokens and the licence apache-2.0.

Facts

Released
2026-07-27 the day the repository was first published on Hugging Face Hugging Face, read
Licence
apache-2.0 the licence the model card declares Hugging Face, read
Open weights
Yes the weights are published in this Hugging Face repository Hugging Face, read
Parameters
266.0 billion counted from the safetensors weight files Hugging Face, read
Active parameters
1.7 billion estimate estimated from the expert counts and sizes in the configuration model configuration, read
Knowledge cutoff
Unknown
Context window
Unknown
Longest output
Unknown
Tool calling
Unknown
Structured output
Unknown
Reasoning controls
Unknown
Inputs and outputs
text, image in, text out Hugging Face, read
Good for
  • Vision: Hugging Face files it under image-text-to-text Hugging Face, read

Prices

Each provider's own price in US dollars per million tokens, as an openly licensed source lists it
ProviderInputOutputContextAs ofSource
Deep Infra0.450 USD1.20 USD524,288 older than 30 days; check the providermodels.dev, read
Kilo Gateway0.450 USD1.20 USD524,288 older than 30 days; check the providermodels.dev, read
NanoGPT0.450 USD1.20 USD524,288 older than 30 days; check the providermodels.dev, read
OpenRouter0.450 USD1.20 USD524,288 older than 30 days; check the providermodels.dev, read
Vercel AI Gateway0.450 USD1.20 USD1,000,000 older than 30 days; check the providermodels.dev, read
Arcee0.500 USD1.20 USD262,144 older than 30 days; check the providermodels.dev, read
Baseten0.500 USD1.20 USD1,048,576 older than 30 days; check the providermodels.dev, read
CommandCode0.500 USD1.20 USD1,000,000 older than 30 days; check the providermodels.dev, read
Hugging Face0.500 USD1.20 USD524,288 older than 30 days; check the providermodels.dev, read
Pioneer0.500 USD1.20 USD1,000,000 older than 30 days; check the providermodels.dev, read
LLMTR0.580 USD1.44 USD262,144 older than 30 days; check the providermodels.dev, read
AI/ML API0.798 USD1.98 USD524,288 older than 30 days; check the providermodels.dev, read

Run it on your own hardware

Weights are the sizes of the files a source lists, or an estimate from the parameter count where none does. Memory is an estimate: weights plus KV cache plus 512 MiB and 5 percent of the weights for runtime buffers. Check it against your hardware.

QuantizationWeightsMemory at 8,192 tokensMemory at 32,768 tokens
UD-IQ1_S69.6 GiB74.9 GiB78.9 GiB
UD-IQ1_M73.4 GiB78.9 GiB82.8 GiB
UD-IQ2_XXS76.6 GiB82.3 GiB86.2 GiB
UD-IQ2_M76.8 GiB82.4 GiB86.4 GiB
UD-Q2_K_XL81.9 GiB87.8 GiB91.7 GiB
UD-IQ3_XXS91.2 GiB97.6 GiB102 GiB
UD-IQ3_S100 GiB107 GiB111 GiB
UD-Q3_K_M111 GiB119 GiB122 GiB
UD-Q3_K_XL111 GiB119 GiB123 GiB
UD-IQ4_NL121 GiB129 GiB133 GiB
UD-Q4_K_S142 GiB151 GiB155 GiB
MXFP4_MOE147 GiB156 GiB160 GiB
UD-Q4_K_M151 GiB161 GiB165 GiB
UD-Q4_K_XL152 GiB161 GiB165 GiB
UD-Q5_K_S172 GiB182 GiB186 GiB
UD-Q5_K_M183 GiB194 GiB198 GiB
UD-Q5_K_XL183 GiB194 GiB198 GiB
UD-Q6_K204 GiB216 GiB220 GiB
UD-Q6_K_XL223 GiB236 GiB240 GiB
Q8_0261 GiB276 GiB280 GiB
UD-Q8_K_XL269 GiB284 GiB288 GiB
BF16491 GiB518 GiB522 GiB

Start it with a local runtime

  • Ollama
    ollama run hf.co/unsloth/Inkling-Small-GGUF:UD-IQ1_S
  • LM Studio
    lms get unsloth/Inkling-Small-GGUF
    lms server start
  • llama.cpp server
    llama-server -hf unsloth/Inkling-Small-GGUF:UD-IQ1_S --jinja
  • vLLM
    vllm serve thinkingmachines/Inkling-Small
  • SGLang
    sglang serve --model-path thinkingmachines/Inkling-Small --port 30000

Use it from your harness

Set up for AI/ML API with the model thinkingmachines/inkling-small. Each endpoint page has the same setup for its own address.

OpenCode

Put this in opencode.json in your project folder:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "aimlapi": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "AI/ML API",
      "options": {
        "baseURL": "https://api.aimlapi.com/v1",
        "apiKey": "{env:AIMLAPI_API_KEY}"
      },
      "models": {
        "thinkingmachines/inkling-small": {
          "name": "Inkling-Small"
        }
      }
    }
  }
}
  • OpenCode reads any OpenAI-compatible address through the @ai-sdk/openai-compatible package, and an address that speaks the Responses API through @ai-sdk/openai.

From OpenCode documentation, read .

Pi

Put this in ~/.pi/agent/models.json:

{
  "providers": {
    "aimlapi": {
      "baseUrl": "https://api.aimlapi.com/v1",
      "api": "openai-completions",
      "apiKey": "$AIMLAPI_API_KEY",
      "models": [
        {
          "id": "thinkingmachines/inkling-small"
        }
      ]
    }
  }
}
  • The apiKey field can name an environment variable as $NAME.

From Pi documentation, read .

Codex

Codex speaks only the Responses API, and AI/ML API documents no Responses address. A gateway that offers one can sit in between.

  • Codex speaks the Responses API only: responses is the one supported wire API of a custom provider. Ollama and LM Studio are built in and start with --oss.

From Codex documentation, read .

Claude Code

Claude Code sends Anthropic Messages requests, and AI/ML API documents no such address. A gateway that translates to that API can sit in between.

  • Claude Code sends Anthropic Messages requests to ANTHROPIC_BASE_URL. Anthropic says it does not support routing Claude Code to models other than Claude through any gateway, so some features may not work with another model.

From Claude Code documentation, read .

Other ways to reach it

  • Arcee, model thinkingmachines/inkling-small
  • Baseten, model thinkingmachines/inkling-small
  • CommandCode, model thinkingmachines/inkling-small
  • DeepInfra, model thinkingmachines/Inkling-Small
  • Hugging Face, model thinkingmachines/Inkling-Small
  • Kilo Gateway, model thinkingmachines/inkling-small
  • LLMTR, model thinkingmachines/inkling-small
  • NanoGPT, model thinkingmachines/Inkling-Small
  • OpenRouter, model thinkingmachines/inkling-small
  • Pioneer, model thinkingmachines/inkling-small

Published results

These are results other people published. Baltor did not run them and does not rank models by them.

Two more publishers are linked, never copied, because their terms do not allow it:

  • Artificial Analysis publishes its own intelligence, speed and price measurements.
  • OpenRouter lists current prices and uptime for each provider it routes to.