embirolabs
Join waitlist

Models for
your next move.

19 open-weight candidates. Find your fit. Compare capabilities, explore benchmarks, and shape the pilot.

Pilot shortlist · not hosted yet
Modular compute layers representing a choice of open models.

19 of 19 model candidates

Z.aiLarge size

GLM 5.3

Complex coding and agents

Context1M
Parameters753B
InputTextOutputText
TextReasoningCodeFunction calling
Terminal-Bench 3.0 28.3%Publisher result
Explore
QwenLight size

Qwen3-8B

Multilingual assistants

Context32K
Parameters8.2B
InputTextOutputText
TextReasoningCodeFunction calling
Benchmarks not listed yet
Explore
Mistral AILarge size

Mistral Small 4

Documents and agents

Context1M
Parameters119B
InputText + ImageOutputText
TextVisionReasoningCode
Benchmarks not listed yet
Explore
Google DeepMindMid size

Gemma 4 31B

Visual understanding

Context256K
Parameters31B
InputText + ImageOutputText
TextVisionReasoning
Benchmarks not listed yet
Explore
Z.aiLarge size

GLM-4.7

Coding and agentic tasks

Context198K
Parameters355B
InputTextOutputText
TextReasoningCode
Benchmarks not listed yet
Explore
QwenLight size

Qwen2.5-VL 7B Instruct

Forms, scans and charts

Context125K
Parameters7B
InputText + ImageOutputText
VisionText
Benchmarks not listed yet
Explore
QwenLight size

Qwen3 Embedding 0.6B

Semantic search

Context32K
Parameters0.6B
InputTextOutputEmbeddings
Embeddings
Benchmarks not listed yet
Explore
BAAILight size

BGE-M3

Hybrid retrieval

Context8K
Parameters0.568B
InputTextOutputEmbeddings
Embeddings
Benchmarks not listed yet
Explore
QwenLight size

Qwen3 Reranker 0.6B

Search result relevance

Context32K
Parameters0.6B
InputTextOutputScores
Reranking
Benchmarks not listed yet
Explore
OpenAILight size

Whisper large-v3

Speech to text

ContextNot specified
Parameters1.55B
InputAudioOutputText
Audio
Benchmarks not listed yet
Explore

Showing 1–12 of 19

Page 1 of 2

All models are evaluation candidates. No live inference or confirmed pricing yet. Publisher marks identify models, not partnerships.

Missing your model?

Tell us what your product needs. Demand helps shape the first deployment.

Request another model →