Novita AI

Inference host · 30 models · 30 priced routes · novita.ai

Models on Novita AI

About Novita AI

Across GlyphCompany profile blogs socials

Model	In / out per 1M	Vs headline
GLM 4.7 Flash bf16 · As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency.	$0.07 / $0.4+5% vs headline	+5%	bf16 · As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency.
Gemma 4 26B A4B bf16 · Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind.	$0.13 / $0.4+132% vs headline	+132%	bf16 · Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind.
Gemma 4 31B bf16 · Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output.	$0.14 / $0.4+28% vs headline	+28%	bf16 · Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output.
GLM 5.3 Flash fp8 · GLM-5.3-Flash is a native multimodal model from Z.ai.	$0.15 / $0.5+389% vs headline	+389%	fp8 · GLM-5.3-Flash is a native multimodal model from Z.ai.
GLM 4.5 Air bf16 · GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications.	$0.13 / $0.85Same vs headline	Same	bf16 · GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications.
Qwen3 VL 30B A3B Instruct bf16 · Qwen3-VL-30B-A3B-Instruct is a multimodal model that unifies strong text generation with visual understanding for images and videos.	$0.2 / $0.7+24% vs headline	+24%	bf16 · Qwen3-VL-30B-A3B-Instruct is a multimodal model that unifies strong text generation with visual understanding for images and videos.
GLM 4.6V bf16 · GLM-4.6V is a large multimodal model designed for high-fidelity visual understanding and long-context reasoning across images, documents, and mixed media.	$0.3 / $0.9Same vs headline	Same	bf16 · GLM-4.6V is a large multimodal model designed for high-fidelity visual understanding and long-context reasoning across images, documents, and mixed media.
DeepSeek V4.1 Flash fp8 · DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture.	$0.3 / $1.2+383% vs headline	+383%	fp8 · DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture.
DeepSeek V4.1 Flash 1017 fp8 · DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the deepseek-flash API model name.	$0.3 / $1.2+100% vs headline	+100%	fp8 · DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the deepseek-flash API model name.
DeepSeek V4 Flash 0731 fp8 · DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total.	$0.44 / $1.32+639% vs headline	+639%	fp8 · DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total.
Qwen3.5-27B bf16 · The Qwen3.5 27B native vision-language Dense model incorporates a linear attention mechanism, delivering fast response times while balancing inference speed and performance.	$0.3 / $2.4+54% vs headline	+54%	bf16 · The Qwen3.5 27B native vision-language Dense model incorporates a linear attention mechanism, delivering fast response times while balancing inference speed and performance.
GLM 4.5V fp8 · GLM-4.5V is a vision-language foundation model for multimodal agent applications.	$0.6 / $1.8Same vs headline	Same	fp8 · GLM-4.5V is a vision-language foundation model for multimodal agent applications.
GLM 4.6 bf16 · Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model to handle more complex...	$0.55 / $2.2+38% vs headline	+38%	bf16 · Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model to handle more complex...
GLM 4.7 fp8 · GLM-4.7 is Z.ai’s latest flagship model, featuring upgrades in two key areas: enhanced programming capabilities and more stable multi-step reasoning/execution.	$0.6 / $2.2+69% vs headline	+69%	fp8 · GLM-4.7 is Z.ai’s latest flagship model, featuring upgrades in two key areas: enhanced programming capabilities and more stable multi-step reasoning/execution.
Kimi K2 0711 fp8 · Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass.	$0.57 / $2.3Same vs headline	Same	fp8 · Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass.
Qwen3.8 27B Qwen3.8 27B is an open-weight dense vision-language model from Qwen.	$0.42 / $3+11% vs headline	+11%	Qwen3.8 27B is an open-weight dense vision-language model from Qwen.
Kimi K2 Thinking bf16 · Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning.	$0.6 / $2.5Same vs headline	Same	bf16 · Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning.
Kimi K2 0905 fp8 · Kimi K2 0905 is the September update of Kimi K2 0711.	$0.6 / $2.5Same vs headline	Same	fp8 · Kimi K2 0905 is the September update of Kimi K2 0711.
Qwen3.5-122B-A10B bf16 · The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency.	$0.4 / $3.2+54% vs headline	+54%	bf16 · The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency.
Kimi K2.5 Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm.	$0.6 / $3+21% vs headline	+21%	Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm.