fireworks/capabilities

9 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

Fireworks reasoning model middleware

Fireworks exposes thinking in deepseek-r1 models using the `<think>` tag. Use the `extractReasoningMiddleware` with `wrapLanguageModel` to extract reasoning and expose it as a `reasoning` property on the result.

Fireworks language model capabilities table

Fireworks language model capabilities: accounts/fireworks/models/firefunction-v1 (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: Yes), accounts/fireworks/models/deepseek-r1 (Image Input: No, Object Generation: Yes, Tool Usage: No, Tool Streaming: No), accounts/fireworks/models/deepseek-v3 (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/llama-v3p1-405b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: Yes), accounts/fireworks/models/llama-v3p1-8b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/llama-v3p2-3b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/llama-v3p3-70b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/mixtral-8x7b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/mixtral-8x7b-instruct-hf (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/mixtral-8x22b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/qwen2p5-coder-32b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/qwen2p5-72b-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/qwen-qwq-32b-preview (Image Input: No, Object Generation: Yes, Tool Usage: No, Tool Streaming: No), accounts/fireworks/models/qwen2-vl-72b-instruct (Image Input: Yes, Object Generation: Yes, Tool Usage: No, Tool Streaming: No), accounts/fireworks/models/llama-v3p2-11b-vision-instruct (Image Input: Yes, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/qwq-32b (Image Input: No, Object Generation: Yes, Tool Usage: No, Tool Streaming: No), accounts/fireworks/models/yi-large (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/kimi-k2-instruct (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/kimi-k2-thinking (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/kimi-k2p6 (Image Input: Yes, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No), accounts/fireworks/models/minimax-m2 (Image Input: No, Object Generation: Yes, Tool Usage: Yes, Tool Streaming: No).

Fireworks embedding model capabilities

Fireworks embedding model capabilities: nomic-ai/nomic-embed-text-v1.5 (Dimensions: 768, Max Tokens: 8192).

Fireworks image editing with Kontext models

Fireworks supports image editing through FLUX Kontext models (flux-kontext-pro and flux-kontext-max). Pass input images via `prompt.images` to transform or edit existing images. Kontext models do not support explicit masks; editing is prompt-driven.

Fireworks image input formats

Input images for Fireworks image editing can be provided as Buffer, ArrayBuffer, Uint8Array, or base64-encoded strings. Fireworks only supports a single input image per request.

Fireworks image model aspect ratios

For all Fireworks image models supporting aspect ratios, the following aspect ratios are supported: 1:1 (default), 2:3, 3:2, 4:5, 5:4, 16:9, 9:16, 9:21, 21:9.

Fireworks image model sizes

For all Fireworks image models supporting size, the following sizes are supported: 640 x 1536, 768 x 1344, 832 x 1216, 896 x 1152, 1024x1024 (default), 1152 x 896, 1216 x 832, 1344 x 768, 1536 x 640.

Fireworks image model capabilities table

Fireworks image model capabilities: accounts/fireworks/models/flux-kontext-pro (Dimensions Specification: Aspect Ratio, Image Editing: Yes), accounts/fireworks/models/flux-kontext-max (Dimensions Specification: Aspect Ratio, Image Editing: Yes), accounts/fireworks/models/flux-1-dev-fp8 (Dimensions Specification: Aspect Ratio, Image Editing: No), accounts/fireworks/models/flux-1-schnell-fp8 (Dimensions Specification: Aspect Ratio, Image Editing: No), accounts/fireworks/models/playground-v2-5-1024px-aesthetic (Dimensions Specification: Size, Image Editing: No), accounts/fireworks/models/japanese-stable-diffusion-xl (Dimensions Specification: Size, Image Editing: No), accounts/fireworks/models/playground-v2-1024px-aesthetic (Dimensions Specification: Size, Image Editing: No), accounts/fireworks/models/SSD-1B (Dimensions Specification: Size, Image Editing: No), accounts/fireworks/models/stable-diffusion-xl-1024-v1-0 (Dimensions Specification: Size, Image Editing: No).

Fireworks models support matrix

Fireworks models and their capabilities: accounts/fireworks/models/deepseek-r1 does not support Image Input, Object Generation, Tool Usage, or Tool Streaming; accounts/fireworks/models/deepseek-v3 does not support Image Input but supports Object Generation and Tool Usage but not Tool Streaming; accounts/fireworks/models/llama-v3p3-70b-instruct does not support Image Input but supports Object Generation, Tool Usage, and Tool Streaming; accounts/fireworks/models/qwen2-vl-72b-instruct supports Image Input but does not support Object Generation, Tool Usage, or Tool Streaming.

Give your agent this brain