EXPLORE / MODEL DIRECTORY

Explore models.Know what runs.

Review 49 open-weight and coding models curated by Awesome Free Models. A listing is a research reference—not a promise that the model is callable through Kaista Cloud.

49Open-model referencesThird-party catalog · CC0

OPEN-MODEL REFERENCE LIBRARY

All open-model references

49 results · Page 2

open-weight

NousResearchOPEN REFERENCE

Hermes 4 (NousResearch)

REFERENCE · awesome-free-models/hermes-4

Feb 2026. Self-improving agentic model with closed-loop learning. Curates own memory and builds skills from experience. Apache 2.0.

Request model review →Hugging Face ↗

open-weight

SnowflakeOPEN REFERENCE

Snowflake Arctic

REFERENCE · awesome-free-models/snowflake-arctic

Apr 2024. Enterprise MoE model balancing high-quality performance with efficient training costs. Optimized for complex data operations. Apache 2.0.

Request model review →Hugging Face ↗

open-weight

TIIOPEN REFERENCE

Falcon 3 (TII)

REFERENCE · awesome-free-models/falcon-3

Dec 2024. Compact high-performance model with strong reasoning. Designed for efficient deployment on resource-constrained hardware. TII Falcon-LLM License 2.0.

Request model review →Hugging Face ↗

open-weight

AppleOPEN REFERENCE

Apple OpenELM

REFERENCE · awesome-free-models/apple-openelm

Apr 2024. Family of efficient on-device SLMs using layer-wise attention scaling. Runs locally on Apple Silicon with full privacy. Apple Sample Code License.

Request model review →Hugging Face ↗

code

MicrosoftOPEN REFERENCE

MAI-Code-1-Flash (Microsoft)

REFERENCE · awesome-free-models/mai-code-1-flash

Jun 2026. Microsoft's open-weight coding model for lowering infrastructure costs.

Request model review →Hugging Face ↗

code

DeepSeekOPEN REFERENCE

DeepSeek Coder

REFERENCE · awesome-free-models/deepseek-coder

State-of-the-art open-weight code generation. DeepSeek's coder series leads SWE-bench. MIT license.

Request model review →Hugging Face ↗

code

AlibabaOPEN REFERENCE

Qwen2.5-Coder (Alibaba)

REFERENCE · awesome-free-models/qwen2-5-coder

Highly capable code model series (1.5B–32B). Excellent balance of speed and quality. Apache 2.0.

Request model review →Hugging Face ↗

code

MistralOPEN REFERENCE

Codestral (Mistral)

REFERENCE · awesome-free-models/codestral

Mistral's dedicated code generation model — fill-in-the-middle, completion, and instruction.

Request model review →Hugging Face ↗

code

GoogleOPEN REFERENCE

CodeGemma (Google)

REFERENCE · awesome-free-models/codegemma

Google's Gemma architecture fine-tuned for code completion and instruction. Apache 2.0.

Request model review →Hugging Face ↗

code

BigCodeOPEN REFERENCE

StarCoder2 (BigCode)

REFERENCE · awesome-free-models/starcoder2

Transparently trained code model covering 619 languages. OpenRAIL-M license.

Request model review →Hugging Face ↗

code

01.AIOPEN REFERENCE

Yi-Coder (01.AI)

REFERENCE · awesome-free-models/yi-coder

Efficient coding model with strong long-context understanding. Yi License (Apache 2.0 compatible).

Request model review →Hugging Face ↗

code

IBMOPEN REFERENCE

Granite Code (IBM)

REFERENCE · awesome-free-models/granite-code

IBM's enterprise-grade code model, available in multiple sizes. Apache 2.0.

Request model review →Hugging Face ↗

code

MicrosoftOPEN REFERENCE

Phi-4-mini (Microsoft)

REFERENCE · awesome-free-models/phi-4-mini

Lightweight model optimized for reasoning and code. Punches above its weight class. MIT license.

Request model review →Hugging Face ↗

code

AlibabaOPEN REFERENCE

Qwen3-Coder-Next (Alibaba)

REFERENCE · awesome-free-models/qwen3-coder-next

Early 2026. Latest generation of Qwen's code series. Strong reasoning and long-context coding capabilities. Apache 2.0.

Request model review →Hugging Face ↗

code

MetaOPEN REFERENCE

CodeLlama (Meta)

REFERENCE · awesome-free-models/codellama

Aug 2023. Llama 2-based code generation pioneer. Supports infilling, completion, and instruction. Llama 2 Community License.

Request model review →Hugging Face ↗

code

WizardLMOPEN REFERENCE

WizardCoder (WizardLM)

REFERENCE · awesome-free-models/wizardcoder

2023. Evol-Instruct fine-tuned for complex coding tasks. Strong general code generation performance. Apache 2.0.

Request model review →Hugging Face ↗

code

m-a-pOPEN REFERENCE

OpenCodeInterpreter

REFERENCE · awesome-free-models/opencodeinterpreter

2024. Integrates execution feedback to iteratively improve generated code. Bridges generation and execution. Apache 2.0.

Request model review →Hugging Face ↗

code

Stability AIOPEN REFERENCE

Stable Code 3B (Stability AI)

REFERENCE · awesome-free-models/stable-code-3b

Aug 2023. Lightweight 3B code model optimized for fill-in-the-middle. Efficient for local autocompletion. StabilityAI license.

Request model review →Hugging Face ↗

code

THUDMOPEN REFERENCE

CodeGeeX2 (THUDM)

REFERENCE · awesome-free-models/codegeex2

2023. Multilingual code model supporting 20+ languages. Strong in both Chinese and English code tasks. Apache 2.0.

Request model review →Hugging Face ↗

code

SalesforceOPEN REFERENCE

CodeT5+ (Salesforce)

REFERENCE · awesome-free-models/codet5

2023. Encoder-decoder architecture unifying code generation, completion, and understanding. BSD-3 license.

Request model review →Hugging Face ↗

code

BigCodeOPEN REFERENCE

SantaCoder (BigCode)

REFERENCE · awesome-free-models/santacoder

2023. Light 1.1B model specialized for Python, Java, and JavaScript. Fast and efficient for IDE integration.

Request model review →Hugging Face ↗