Speka
Start free
Catalog

6 models, one API key

Access DeepSeek V4, Nemotron 3, Llama, Mistral, GPT-OSS, FLUX and more frontier models through one OpenAI-compatible API key — every model priced per token, no subscriptions. Every model below is health-checked continuously, so the catalog only lists what is actually running. Filter by capability, copy the model id, and start calling Speka in minutes.

What models can I call on Speka?

Speka is a unified AI model API that provides developer access to 6 frontier open-weight models — including DeepSeek V4, Nemotron 3 Ultra 550B, Mistral Nemotron, GPT-OSS and FLUX.1 for image generation — through a single OpenAI-compatible endpoint. Pricing is transparent and per-token, with no subscriptions or seat fees. Input prices run from $0.01 per million tokens for embeddings to $0.90 per million tokens for Nemotron 3 Ultra 550B. Developers switch between models by changing one parameter — the model field — with no SDK changes required. Model weights and benchmarks for these open models are published by their labs — for example DeepSeek and Black Forest Labs (FLUX) on Hugging Face.

O

GPT-OSS 20B

OpenAI
code

The small GPT-OSS tier — quick code completion, refactors and shell/tool calls at a fraction of the price.

openai/gpt-oss-20b
Input / 1M
$0.07
Output / 1M
$0.30
Open model
NVIDIA logo

Nemotron 3 Embed 1B

NVIDIA
embedding

Nemotron 3 embedding model for semantic search, clustering and reranking-style similarity scoring.

nvidia/nemotron-3-embed-1b
Input / 1M
$0.012
Open model
DeepSeek logo

Deepseek v4 Flash 0731

DeepSeek
reasoning

Auto-discovered from the NVIDIA catalog on 2026-08-07.

deepseek-ai/deepseek-v4-flash-0731
Input / 1M
$0.20
Output / 1M
$0.20
Open model
DeepSeek logo

Deepseek v4 Pro 0813

DeepSeek
reasoning

Auto-discovered from the NVIDIA catalog on 2026-08-26.

deepseek-ai/deepseek-v4-pro-0813
Input / 1M
$0.20
Output / 1M
$0.20
Open model
NVIDIA logo

Nemotron 3 Super 120B

NVIDIA
reasoning

NVIDIA's mixture-of-experts reasoning model — 120B total parameters with 12B active, so frontier-class reasoning arrives at small-model latency.

nvidia/nemotron-3-super-120b-a12b
Input / 1M
$0.40
Output / 1M
$0.60
Open model
Meta logo

Llama 3.2 11B Vision

Meta
vision

Lightweight vision-language model for fast image captioning, OCR and visual Q&A.

meta/llama-3.2-11b-vision-instruct
Input / 1M
$0.06
Output / 1M
$0.06
Open model
FAQ

Frequently asked questions

Speka provides 6 frontier open-weight models across six categories: reasoning (DeepSeek V4 Flash, DeepSeek V4 Pro, Nemotron 3 Super 120B, Nemotron 3 Nano 30B, Nemotron Super 49B, Nemotron Nano 9B), chat (Nemotron 3 Ultra 550B, Llama 3.3 70B, Llama 3.1 8B, Llama 3.2 3B, Mistral Nemotron, Mistral Medium 3.5, GLM 5.2, MiniMax M3, Step 3.7 Flash), code (GPT-OSS 120B, GPT-OSS 20B, Laguna XS 2.1), vision (Llama 3.2 90B Vision, Nemotron Nano 12B VL, Llama 3.2 11B Vision), embeddings (NV-EmbedQA E5 v5, Llama Nemotron Embed 1B v2, Nemotron 3 Embed 1B, NV-Embed v1), and image generation (FLUX.1 [dev], FLUX.1 [schnell]). The live catalog is verified continuously — see the model list below for what is running right now.
DeepSeek V4 Flash on Speka costs $0.27 per million input tokens and $1.10 per million output tokens. Pricing is per-token with no subscription or seat fee — you pay only for what you use, with no minimum commitment.
Yes. Speka uses an OpenAI-compatible endpoint, so any application already using the OpenAI Python SDK, TypeScript SDK, or HTTP API can switch to Speka by changing two values: the base URL and the model ID. No other code changes are required.
GPT-OSS 120B is the flagship code model in the catalog — a 120-billion-parameter open-weight model strong at code generation, refactoring, and tool-use, priced at $0.15 per million input tokens and $0.60 per million output tokens. GPT-OSS 20B ($0.07 / $0.30 per million tokens) and Poolside Laguna XS 2.1 cover latency-sensitive completion, and Llama 3.1 8B Instruct at $0.05 per million tokens is the cheapest capable alternative.
Yes. Speka includes four embedding models: NV-EmbedQA E5 v5, tuned for question-answering retrieval at $0.01 per million tokens; Llama Nemotron Embed 1B v2 and Nemotron 3 Embed 1B, current-generation multilingual retrieval models at $0.012 per million tokens; and NV-Embed v1, a general-purpose model at $0.016 per million tokens. All return standard embedding vectors compatible with any vector database. For the generation step, any Speka chat or reasoning model is available under the same API key.