Skip to content

Supported models

Every training run names two models. Use only the values on this page: a student_model_name or teacher_model_name from outside this catalog fails validation.

  • Student. The small model fine-tuned for your task. This is the one you deploy, so pick a size that fits your latency, cost and hardware constraints.
  • Teacher. The large model that writes and checks your synthetic training data. Its knowledge is distilled into the student, and you never deploy it.
Section titled “Recommended models”

Student: the Qwen3.5 family. Start with Qwen3.5-4B. Go smaller (Qwen3.5-2B, Qwen3.5-0.8B) when your deployment target needs it, or larger (Qwen3.5-9B) when accuracy matters more than cost.

Teachers: GLM 5.3. zai.glm-5.3-flash-low-thinking is the small teacher, for simple tasks and for cost. zai.glm-5.3-low-thinking is the large teacher, for everything else. Move either to its -high-thinking value when teacher evaluation shows the task needs more reasoning.

The judge. evaluation.llm_as_a_judge_model_name scores every run, and scores from two judges aren’t comparable, so it is usually set in the first config and kept (Variation between runs). When it’s omitted, it takes the value of base.teacher_model_name at the time the job is created and is written into the config, so a config read back for an override names it, and changing the teacher there leaves the judge as it is. trace_processing.teacher_model_name, the relabeling model, works the same way.

When the fields are omitted, the config defaults are Llama-3.2-1B-Instruct and nvidia.nemotron-3.5-super-vl. A config with the recommended models:

base:
  task: question-answering
  student_model_name: Qwen3.5-4B
  teacher_model_name: zai.glm-5.3-low-thinking
evaluation:
  llm_as_a_judge_model_name: zai.glm-5.3-low-thinking

Student models

Section titled “Student models”
Family Values Tool calling
Qwen3.5 Qwen3.5-0.8B, Qwen3.5-2B, Qwen3.5-4B, Qwen3.5-9B ✓
Llama 3 Llama-3.2-1B-Instruct, Llama-3.2-3B-Instruct, Llama-3.1-8B-Instruct ✓
Qwen3 Qwen3-0.6B, Qwen3-1.7B, Qwen3-4B-Instruct-2507, Qwen3-8B ✓
LFM2 / LFM2.5 LFM2-350M, LFM2-1.2B, LFM2-2.6B, LFM2.5-350M, LFM2.5-1.2B-Instruct ✓
Gemma 4 gemma-4-E2B-it, gemma-4-E4B-it ✓
Qwen3.6 Qwen3.6-35B-A3B ✓
Qwen3.8 Qwen3.8-27B (QLoRA only) ✓
FunctionGemma functiongemma-270m-it ✓
Granite 4.2 Granite-4.2-8B ✓
Nemotron 3.5 Nemotron-3.5-Lightning-30B-A3B (no QLoRA) ✓
Gemma 3 gemma-3-270m-it, gemma-3-1b-it, gemma-3-4b-it ✗
SmolLM2 SmolLM2-135M-Instruct, SmolLM2-1.7B-Instruct ✗

Qwen3.8-27B can only be trained with tuning.use_qlora: true. Config validation doesn’t check this, so set it in every config that trains it. Nemotron-3.5-Lightning-30B-A3B can’t be trained with QLoRA: config validation rejects tuning.use_qlora: true for it.

Size tiers

Section titled “Size tiers”
Size range When to choose
135M - 350M Extreme latency constraints, on-device, very simple tasks
0.6B - 1.2B Cost-sensitive production, well-defined tasks
1.7B - 3B Better quality than the 1B class within a small memory and latency budget
4B The default: start here
8B - 9B Accuracy-critical tasks, multi-step reasoning, nuanced outputs
27B dense Highest quality from a dense model; needs a large GPU to serve
30B - 35B MoE (3B active) Highest quality; inference costs as much as a 3B model, but the weights need the memory of a 35B one

Trying several sizes means one training run per student against the same dataset. See Model training.

Teacher models

Section titled “Teacher models”

A teacher is open-weight or proprietary. Distill only from an open-weight teacher. Proprietary models are not suitable for distillation: the platform accepts them as base.teacher_model_name, but use them only in a teacher evaluation, to score the production model the student replaces.

Family Values Distillation
GLM zai.glm-5.3-low-thinking, zai.glm-5.3-high-thinking, zai.glm-5.3-flash-low-thinking, zai.glm-5.3-flash-high-thinking, zai.glm-5, zai.glm-5-thinking, zai.glm-5.2, zai.glm-5.2-thinking ✓
GPT OSS openai.gpt-oss-20b, openai.gpt-oss-20b-thinking, openai.gpt-oss-120b, openai.gpt-oss-120b-thinking ✓
DeepSeek deepseek.v3.1, deepseek.v4-pro, deepseek.v4-pro-thinking, deepseek.v4-pro-0813-low-thinking, deepseek.v4-pro-0813-high-thinking, deepseek.v4.1-flash-low-thinking, deepseek.v4.1-flash-high-thinking, deepseek.v4-flash-vision-exp, deepseek.v4-flash-vision-exp-thinking ✓
Qwen Qwen3-235B-A22B-Instruct-2507, Qwen3-480B-A35B-Coder, Qwen2.5-VL-72B-Instruct, Qwen3.8-2.4T-A95B-minimal-thinking, Qwen3.8-2.4T-A95B-medium-thinking, qwen.qwen3.8-flash, qwen.qwen3.8-flash-thinking, qwen.qwen3.8-27b, qwen.qwen3.8-27b-thinking ✓
Kimi moonshotai.kimi-k2.6, moonshotai.kimi-k2.6-thinking, moonshotai.kimi-k3, moonshotai.kimi-k3-low-thinking, moonshotai.kimi-k3-max-thinking ✓
Nemotron nvidia.nemotron-3.5-super-vl, nvidia.nemotron-3.5-super-vl-thinking, nvidia.nemotron-3-ultra ✓
Anthropic Claude (proprietary) anthropic.claude-opus-4-6, anthropic.claude-opus-5, anthropic.claude-opus-5-thinking, anthropic.claude-sonnet-5, anthropic.claude-opus-5-5, anthropic.claude-opus-5-5-thinking, anthropic.claude-sonnet-5-5, anthropic.claude-sonnet-5-5-thinking ✗ teacher evaluation only
OpenAI GPT-5 / GPT-6 (proprietary) openai.gpt-5-mini, openai.gpt-5.1, openai.gpt-5.1-thinking, openai.gpt-5.2, openai.gpt-5.2-thinking, openai.gpt-5.6-sol, openai.gpt-5.6-sol-thinking, openai.gpt-5.6-luna, openai.gpt-5.6-luna-thinking, openai.gpt-5.6-terra, openai.gpt-5.6-terra-thinking, openai.gpt-6-luna, openai.gpt-6-luna-thinking, openai.gpt-6-sol, openai.gpt-6-sol-thinking, openai.gpt-6.1-sol, openai.gpt-6.1-sol-thinking ✗ teacher evaluation only
Google Gemini (proprietary) google.gemini-3.1-pro-preview-thinking, google.gemini-3.1-flash-lite, google.gemini-3.8-flash, google.gemini-3.8-flash-thinking, google.gemini-3.7-flash, google.gemini-3.7-flash-thinking, google.gemini-3.6-flash, google.gemini-3.6-flash-thinking, google.gemini-3.5-flash, google.gemini-3.5-flash-thinking, google.gemini-3.5-flash-lite, google.gemini-3.5-flash-lite-thinking ✗ teacher evaluation only

A plain value and its -thinking value point at the same model: the -thinking value turns the reasoning mode on, the plain value turns it off. Some families name the reasoning effort instead: -low-thinking and -high-thinking (GLM 5.3, DeepSeek V4 Pro 0813, DeepSeek V4.1 Flash), -minimal-thinking and -medium-thinking (Qwen3.8), -low-thinking and -max-thinking (Kimi K3). Claude Opus 5.5 and Sonnet 5.5, GPT-6.1 Sol, Gemini 3.5 to 3.8 Flash and Gemini 3.5 Flash Lite always reason: their plain value runs at the lowest effort the model accepts and their -thinking value at high. qwen.qwen3.8-27b-thinking runs at medium effort. A higher effort gives better answers on hard tasks, and is slower and costs more.

openai.gpt-oss-20b and openai.gpt-oss-120b run at low reasoning effort; their -thinking values run at medium.

Every teacher is a reasoning model except Qwen3-235B-A22B-Instruct-2507, Qwen3-480B-A35B-Coder and Qwen2.5-VL-72B-Instruct. A reasoning teacher needs synthgen.teacher_temperature between 0.5 and 0.7 (Config file).

Reasoning students

Section titled “Reasoning students”

The students that support base.enable_thinking: Reasoning models.

Vision

Section titled “Vision”

base.visual_task: true needs a vision-capable model in every teacher and judge role: base.teacher_model_name, trace_processing.teacher_model_name, evaluation.llm_as_a_judge_model_name and every model in trace_processing.relabelling_committee_models. One non-vision model in any role fails validation.

  • Vision teachers: moonshotai.kimi-k2.6, moonshotai.kimi-k2.6-thinking, moonshotai.kimi-k3, moonshotai.kimi-k3-low-thinking, moonshotai.kimi-k3-max-thinking, zai.glm-5.3-flash-low-thinking, zai.glm-5.3-flash-high-thinking, deepseek.v4.1-flash-low-thinking, deepseek.v4.1-flash-high-thinking, deepseek.v4-flash-vision-exp, deepseek.v4-flash-vision-exp-thinking, Qwen2.5-VL-72B-Instruct, qwen.qwen3.8-flash, qwen.qwen3.8-flash-thinking, qwen.qwen3.8-27b, qwen.qwen3.8-27b-thinking, nvidia.nemotron-3.5-super-vl, nvidia.nemotron-3.5-super-vl-thinking. Every proprietary teacher is also vision-capable, for teacher evaluation only.
  • Vision students: Qwen3.5-0.8B, Qwen3.5-2B, Qwen3.5-4B, Qwen3.5-9B, Qwen3.8-27B, gemma-4-E2B-it, gemma-4-E4B-it.

Tool-calling compatibility

Section titled “Tool-calling compatibility”

This applies to the chat completion tasks whenever the job description declares tools (always, for chat-completion-agentic). A chat-completion job with no tools has no model restriction.

  • Students: only the families marked ✓ in the student table: Qwen3.5, Llama 3, Qwen3, LFM2/LFM2.5, Gemma 4, Qwen3.6, Qwen3.8, FunctionGemma, Granite 4.2 and Nemotron 3.5.
  • Teachers: every teacher except deepseek.v3.1, Qwen3-480B-A35B-Coder and Qwen2.5-VL-72B-Instruct.

Next

Section titled “Next”

Every other parameter and its default: Config file.