Fireworks

53 indexed models

53 matching models

GLM 5 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-GLM-5@1

GLM 5 is a 744B-parameter Mixture-of-Experts model targeting complex systems engineering and long-horizon agentic tasks, using DeepSeek Sparse Attention for efficient long-context processing.

Context window
202,752
Input modalities
text
Output modalities
text
Capabilities
toolCalling

Inkling · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Inkling@1

Inkling is a 975B-parameter sparse Mixture-of-Experts model from Thinking Machines Lab with 41B active parameters, controllable thinking effort, and a 1M-token context window.

Context window
1,048,576
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Kimi K2.5 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K2.5@1

Kimi K2.5 is an open-source, native multimodal agentic model built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base.

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Kimi K2.6 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K2.6@1

Kimi K2.6 is an open-source, native multimodal agentic model with 1T total parameters and 32B activated, featuring long-horizon coding, coding-driven design, elevated agent swarm, and proactive autonomous execution.

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Kimi K2.7 Code · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K2.7-Code@1

Kimi K2.7 Code is a coding-focused agentic model built upon Kimi K2.6, improving real-world long-horizon coding and end-to-end software engineering workflows while reducing thinking-token usage by approximately 30%.

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Kimi K2 Instruct · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K2-Instruct-0905@1

Kimi K2 Instruct is an updated Mixture-of-Experts model with 1 trillion total parameters featuring improved coding abilities, agentic tool use, and an extended 256K token context window.

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
toolCalling

Kimi K2 Thinking · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K2-Thinking@1

Kimi K2 Thinking is an open-source thinking model that reasons step-by-step while dynamically invoking tools, with native INT4 quantization for lossless reductions in inference latency and memory usage.

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Kimi K3.1 Pro · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K3.1-Pro@1

Kimi K3.1 Pro is Moonshot AI's open-weight multimodal agentic model with a 1M-token context window, designed for coding, visual workflows, and long-context research.

Context window
1,048,576
Input modalities
text, image
Output modalities
text
Capabilities
multimodal, reasoning, toolCalling, visionInput

Kimi K3 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Kimi-K3@1

Kimi K3 is Moonshot AI's 2.8T-parameter native-vision model with a 1M-token context window, designed for long-horizon coding, knowledge work, and reasoning.

Context window
1,048,576
Input modalities
text, image
Output modalities
text
Capabilities
multimodal, reasoning, toolCalling, visionInput

Llama 3.1 8B Instruct · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Llama-v3.1-8B-Instruct@1

Llama 3.1 8B Instruct is an 8B-parameter multilingual instruction-tuned language model optimized for dialogue, with a 128K-token context window, tool-calling support, and multilingual capabilities across 8 languages.

Context window
131,072
Input modalities
text
Output modalities
text
Capabilities
toolCalling

MiniMax M2.5 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-MiniMax-M2.5@1

MiniMax M2.5 is a Mixture-of-Experts model built for state-of-the-art coding, agentic tool use, and search, trained with reinforcement learning across hundreds of thousands of real-world environments.

Context window
196,608
Input modalities
text
Output modalities
text
Capabilities
toolCalling

MiniMax M3 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-MiniMax-M3@1

MiniMax M3 is a native multimodal Mixture-of-Experts model from MiniMax with 1M context, about 428B total parameters, about 23B activated parameters, MiniMax Sparse Attention, and strong coding and cowork capabilities.

Context window
1,048,576
Input modalities
text, image
Output modalities
text
Capabilities
multimodal, reasoning, toolCalling, visionInput

Ministral 3 3B Instruct 2512 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Ministral-3-3B-Instruct-2512@1

An FP8 instruct post-trained multimodal model that pairs a 3.4B language model with a 0.4B vision encoder, delivering multilingual chat, image analysis, native function calling and JSON output, and a 256K context window for edge deployment.

Context window
262,144
Input modalities
text, image
Output modalities
text
Capabilities
multimodal, toolCalling, visionInput

Muse Glimmer 30B · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Muse-Glimmer-30B@1

Muse Glimmer 30B is Meta's dense causal language model distilled from Muse Spark for autonomous agentic work, listed by Fireworks with 29.7B parameters, text and image input, a 131K-token context window, schema-based tool calling, failure recovery, and sel

Context window
131,072
Input modalities
text, image
Output modalities
text
Capabilities
multimodal, reasoning, toolCalling, visionInput

Nemotron 3 Ultra NVFP4 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Nemotron-3-Ultra-NVFP4@1

Nemotron 3 Ultra NVFP4 is a frontier-scale Mixture-of-Experts language model from NVIDIA with approximately 550B total parameters and 55B active parameters, a hybrid Mamba-2 and LatentMoE architecture, configurable reasoning, function calling, and a 262,14

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Nemotron Lightning 3.5 30B A3B · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Nemotron-Lightning-3.5-30B-A3B@1

Nemotron Lightning 3.5 30B A3B is a 30B-parameter Mixture-of-Experts language model from NVIDIA with 3B active parameters, a hybrid Mamba-Transformer architecture, configurable reasoning, Multi-Token Prediction, function calling, and a 262,144-token Firewo

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

Nemotron Super 120B · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-Nemotron-3-Super-120B-A12B-BF16@1

Nemotron Super 120B is a 120B-parameter Mixture-of-Experts language model from NVIDIA with 12B active parameters, a hybrid Mamba-2 and LatentMoE architecture, configurable reasoning mode, multilingual support, and a 262,144-token context window — optimized

Context window
262,144
Input modalities
text
Output modalities
text
Capabilities
reasoning, toolCalling

NVIDIA Nemotron 3.5 Super VL 120B A12B NVFP4 · azureml-fireworks v1

azureFoundry/azureml-fireworks/FW-NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-NVFP4@1

NVIDIA Nemotron 3.5 Super VL 120B A12B NVFP4 is an open multimodal model for long-running agents with text, image, document, and video understanding, a hybrid Mamba-Transformer Mixture-of-Experts architecture, 120B total and 12B active parameters, and a 1,

Context window
1,048,576
Input modalities
text, image, video
Output modalities
text
Capabilities
multimodal, reasoning, toolCalling, visionInput

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace