SiliconFlow: Models Intelligence, Performance & Price

Name: Intelligence
Creator: Artificial Analysis
License: https://artificialanalysis.ai/docs/legal/Terms-of-Use.pdf

SiliconFlow

This analysis is intended to support you in choosing the best model provided by SiliconFlow for your use-case.

Most Intelligent

Updated

GLM-5.3-Flash

DeepSeek V4 Pro 0813 (max)

DeepSeek V4 Flash 0731 (max)

GLM-5.2 (max) (FP8)

DeepSeek V4 Pro (max) (FP8)

Intelligence index

Total 35 models

Fastest

MiniMax-M3 (FP8)

166 t/s

DeepSeek V4 Flash 0731 (max)

113 t/s

DeepSeek V4 Flash (high) (FP8)

112 t/s

DeepSeek V4 Flash (max) (FP8)

111 t/s

Gemma 4 12B

110 t/s

Output speed

Total 35 models

Lowest Price

DeepSeek V4 Flash (high) (FP8)

$0.07

DeepSeek V4 Flash (max) (FP8)

$0.07

GLM-5.3-Flash

$0.10

Hy3 (FP8)

$0.10

Qwen3.5 9B (FP8)

$0.11

Blended price (per 1M tokens)

Total 35 models

SiliconFlow offers 35 models, each with different intelligence, performance, and pricing characteristics. Below is a comparison of the key metrics across models.

For intelligence, the top models on SiliconFlow are GLM-5.3-Flash (42), DeepSeek V4 Pro 0813 (max) (36), and DeepSeek V4 Flash 0731 (max) (34).
For output speed, the fastest models are MiniMax-M3 (FP8) (166 t/s), DeepSeek V4 Flash 0731 (max) (113 t/s), and DeepSeek V4 Flash (high) (FP8) (112 t/s). Speed varies significantly across models, with a 51% difference between the fastest and slowest.
For latency, Kimi K2.6 (Non-reasoning) (FP8) (1.62s), GLM-5.2 (Non-reasoning) (FP8) (1.98s), and GLM-5.1 (Non-reasoning) (FP8) (2.38s) offer the lowest time to first answer token.
For pricing, DeepSeek V4 Flash (high) (FP8) ($0.07), DeepSeek V4 Flash (max) (FP8) ($0.07), and GLM-5.3-Flash ($0.10) offer the lowest blended prices per 1M tokens.
For context window size, GLM-5.2 (max) (FP8) (1M), DeepSeek V4 Pro (max) (FP8) (1M), and DeepSeek V4 Pro (high) (FP8) (1M) support the largest context windows on SiliconFlow.

Highlights

Updated

Intelligence

Artificial Analysis Intelligence Index · Higher is better

Speed

Output tokens per second · Higher is better

Price

USD per 1M tokens (blended) · Lower is better

Intelligence Evaluations

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v4.3.2 incorporates 10 evaluations: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1

Estimate (independent evaluation forthcoming)

Artificial Analysis Intelligence Index v4.3.2 includes: AA-Briefcase v1.1, GDPval-AA v2.1, AutomationBench-AA, Terminal-Bench 4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Evaluations

Intelligence evaluations measured independently by Artificial Analysis · Higher is better

AA-Briefcase v1.1Updated

Agentic knowledge work, (Elo-500)/2000

GDPval-AA v2.1Updated

Agentic real-world work tasks, (Elo-500)/2000

AutomationBench-AAUpdated

Agentic SaaS workflows

Terminal-Bench 4.0New

Agentic coding & terminal use

SciCode

Coding

Humanity's Last Exam

Reasoning & knowledge

GDP.pdfNew

Professional document reasoning, All-pass

CritPt

Physics reasoning

AA-Omniscience Accuracy

Knowledge

AA-Omniscience Non-Hallucination Rate

1 - hallucination rate

AA-LCR v1.1

Long context reasoning

Harvey LAB-AA

Legal agentic work, criterion pass rate

EnterpriseOps-Gym-AA

Agentic business operations

AA-AnalystAgent

Quantitative analysis on spreadsheets & documents

𝜏³-Banking

Agentic tool use

ITBench-AA

Kubernetes incident root-cause analysis

MMMU-Pro

Visual reasoning

MLCR-AANew

Medical long context reasoning

While model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Context Window

Context window: tokens limit · Higher is better

Larger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Pricing

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Performance Summary

Output Speed vs. Price

Output speed: output tokens per second · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

Speed

Measured by Output Speed (tokens per second)

Output Speed

Output tokens per second · Higher is better

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Latency

Measured by Time (seconds) to First Token

Latency: Time To First Answer Token

Seconds to first answer token received · Accounts for reasoning model 'thinking' time

Time to first answer token received, in seconds, after API request sent. For reasoning models, this includes the 'thinking' time of the model before providing an answer. For models which do not support streaming, this represents time to receive the completion.

End-to-End Response Time

Seconds to output 500 tokens, calculated based on time to first token, 'thinking' time for reasoning models, and output speed

End-to-End Response Time vs. Price

End-to-end response time: end-to-end seconds to output 500 tokens · USD per 1M tokens (blended)

Most attractive quadrant

Pareto line

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).


									Further Analysis
GLM-5.3-Flash	1M	Open	42	$0.29	94	1.65	28.21	21.25	Model Provider
DeepSeek V4 Pro 0813 (max)	1.05M	Open	36	$6.15	77	1.80	34.09	25.84	Model Provider
DeepSeek V4 Flash 0731 (max)	1.05M	Open	34	$0.85	113	2.06	24.08	17.62	Model Provider
GLM-5.2 (max) (FP8)	1.05M	Open	34	$0.85	98	1.96	27.56	20.48	Model Provider
DeepSeek V4 Pro (max) (FP8)	1.05M	Open	30	--	52	1.79	94.71	83.39	Model Provider
DeepSeek V4 Pro (high) (FP8)	1.05M	Open	30*	--	50	1.98	51.83	39.85	Model Provider
MiniMax-M3 (FP8)	1M	Open	29	$0.45	166	1.14	16.22	12.07	Model Provider
GLM-5 (FP8)	200k	Open	28*	--	--	--	--	--	Model Provider
Kimi K2.6 (FP8)	262k	Open	27	$0.54	25	1.64	201.47	179.65	Model Provider
GLM-5.1 (FP8)	205k	Open	26	$1.66	30	2.49	147.94	128.50	Model Provider
DeepSeek V4 Flash (high) (FP8)	1.05M	Open	26*	--	112	1.69	17.30	11.12	Model Provider
Hy3 (FP8)	256k	Open	25	$0.07	87	3.11	31.94	23.06	Model Provider
GLM-5.1 (Non-reasoning) (FP8)	205k	Open	24*	--	29	2.38	19.41	--	Model Provider
DeepSeek V4 Flash (max) (FP8)	1.05M	Open	24	--	111	1.50	56.73	50.71	Model Provider
Kimi K2.6 (Non-reasoning) (FP8)	262k	Open	24*	--	25	1.62	21.43	--	Model Provider
Kimi K2.5 (FP8)	262k	Open	23*	--	25	2.01	138.97	117.21	Model Provider
Qwen3.5 27B (FP8)	262k	Open	23*	--	35	3.06	74.82	57.41	Model Provider
MiniMax-M2.5 (FP8)	197k	Open	23*	--	--	--	--	--	Model Provider
GLM-5.2 (Non-reasoning) (FP8)	1.05M	Open	22*	--	91	1.98	7.48	--	Model Provider
GLM-5 (Non-reasoning) (FP8)	205k	Open	22*	--	--	--	--	--	Model Provider
DeepSeek V3.2 (FP8)	164k	Open	21*	--	22	2.00	116.11	91.28	Model Provider
Qwen3.6 27B (FP8)	262k	Open	21	$0.36	45	3.41	141.24	126.67	Model Provider
Qwen3.5 35B A3B (FP8)	262k	Open	19*	--	58	1.77	44.71	34.35	Model Provider
LongCat 2.0 (FP8)	262k	Open	19	$0.15	57	2.87	46.68	35.05	Model Provider
Gemma 4 31B (FP8)	262k	Open	19*	--	48	3.69	50.43	36.29	Model Provider
Qwen3.5 397B A17B (FP8)	262k	Open	18	$0.31	--	--	--	--	Model Provider
Qwen3.6 35B A3B (FP8)	262k	Open	18	$0.27	106	2.08	57.89	51.07	Model Provider
Step 3.5 Flash (FP8)	262k	Open	17*	--	75	1.77	35.32	26.84	Model Provider
DeepSeek V3.2 (Non-reasoning) (FP8)	164k	Open	16*	--	21	2.58	26.29	--	Model Provider
Qwen3.5 122B A10B (FP8)	262k	Open	16	$0.21	67	1.52	38.61	29.67	Model Provider
Gemma 4 12B	262k	Open	14*	--	110	2.45	25.24	18.23	Model Provider
Gemma 4 31B (Non-reasoning) (FP8)	262k	Open	14*	--	50	3.58	13.65	--	Model Provider
Qwen3.5 9B (FP8)	262k	Open	14*	--	39	2.15	66.27	51.29	Model Provider
Gemma 4 26B A4B (Non-reasoning) (FP8)	262k	Open	13*	--	84	2.42	8.39	--	Model Provider
Seed-OSS-36B-Instruct	262k	Open	12*	--	40	2.94	66.22	50.62	Model Provider
GLM-4.6V	128k	Open	11*	--	--	--	--	--	Model Provider
GLM-4.5-Air	98.3k	Open	11*	--	72	2.51	37.41	27.92	Model Provider
Gemma 4 12B (Non-reasoning)	262k	Open	9*	--	110	2.48	7.04	--	Model Provider
GLM-4.6V (Non-reasoning)	128k	Open	8*	--	--	--	--	--	Model Provider
Ling-flash-2.0	131k	Open	8*	--	8	2.68	63.80	--	Model Provider
Qwen2.5 72B (FP8)	32k	Open	8*	--	29	4.34	21.70	--	Model Provider
ERNIE 4.5 300B A47B	131k	Open	8*	--	--	--	--	--	Model Provider
Ring-flash-2.0	131k	Open	7*	--	--	--	--	--	Model Provider

Key definitions

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Frequently Asked Questions

Common questions about SiliconFlow

SiliconFlow offers 35 models that we track: GLM-5.3-Flash, DeepSeek V4 Pro 0813 (max), DeepSeek V4 Flash 0731 (max), GLM-5.2 (max) (FP8), DeepSeek V4 Pro (max) (FP8), DeepSeek V4 Pro (high) (FP8), MiniMax-M3 (FP8), Kimi K2.6 (FP8), GLM-5.1 (FP8), DeepSeek V4 Flash (high) (FP8), Hy3 (FP8), GLM-5.1 (Non-reasoning) (FP8), DeepSeek V4 Flash (max) (FP8), Kimi K2.6 (Non-reasoning) (FP8), Kimi K2.5 (FP8), Qwen3.5 27B (FP8), GLM-5.2 (Non-reasoning) (FP8), DeepSeek V3.2 (FP8), Qwen3.6 27B (FP8), Qwen3.5 35B A3B (FP8), LongCat 2.0 (FP8), Gemma 4 31B (FP8), Qwen3.6 35B A3B (FP8), Step 3.5 Flash (FP8), DeepSeek V3.2 (Non-reasoning) (FP8), Qwen3.5 122B A10B (FP8), Gemma 4 12B, Gemma 4 31B (Non-reasoning) (FP8), Qwen3.5 9B (FP8), Gemma 4 26B A4B (Non-reasoning) (FP8), Seed-OSS-36B-Instruct, GLM-4.5-Air, Gemma 4 12B (Non-reasoning), Ling-flash-2.0, and Qwen2.5 72B (FP8).

The most intelligent model available on SiliconFlow is GLM-5.3-Flash with an Intelligence Index score of 42.

The fastest model on SiliconFlow by output speed is MiniMax-M3 (FP8) at 165.8 tokens per second.

The model with the lowest time to first answer token on SiliconFlow is Kimi K2.6 (Non-reasoning) (FP8) at 1.62s. Lower latency means faster initial response time.

The most affordable model on SiliconFlow by blended price is DeepSeek V4 Flash (high) (FP8) at $0.07 per 1M tokens (7:2:1 cache hit/input/output ratio).

Prices on SiliconFlow vary up to 14x across models, from $0.07 per 1M tokens for DeepSeek V4 Flash (high) (FP8) to $1.03 per 1M tokens for GLM-5.1 (Non-reasoning) (FP8).

Yes, SiliconFlow offers an OpenAI-compatible API, making it easy to switch from OpenAI or use existing OpenAI SDK integrations.

28 of 35 models on SiliconFlow support JSON mode for structured output.

Yes, all 35 models on SiliconFlow support function calling (tool use).

Yes, SiliconFlow offers 26 reasoning models: GLM-5.3-Flash, DeepSeek V4 Pro 0813 (max), DeepSeek V4 Flash 0731 (max), GLM-5.2 (max) (FP8), DeepSeek V4 Pro (max) (FP8), DeepSeek V4 Pro (high) (FP8), MiniMax-M3 (FP8), Kimi K2.6 (FP8), GLM-5.1 (FP8), DeepSeek V4 Flash (high) (FP8), Hy3 (FP8), DeepSeek V4 Flash (max) (FP8), Kimi K2.5 (FP8), Qwen3.5 27B (FP8), DeepSeek V3.2 (FP8), Qwen3.6 27B (FP8), Qwen3.5 35B A3B (FP8), LongCat 2.0 (FP8), Gemma 4 31B (FP8), Qwen3.6 35B A3B (FP8), Step 3.5 Flash (FP8), Qwen3.5 122B A10B (FP8), Gemma 4 12B, Qwen3.5 9B (FP8), Seed-OSS-36B-Instruct, and GLM-4.5-Air. Reasoning models use extended thinking to work through complex problems before providing an answer.

Yes, all 35 models on SiliconFlow are open weight models.

Yes, provider performance can vary over time due to infrastructure changes, load balancing, and updates. We continuously benchmark all providers and display historical performance trends in the "Over Time" charts.

When choosing a model on SiliconFlow, consider: intelligence (for quality-sensitive tasks), output speed (for throughput-intensive tasks), latency (for interactive applications requiring quick first responses), pricing (for cost-sensitive workloads), and features like context window size, JSON mode, or function calling support.

SiliconFlow: Models Intelligence, Performance & Price

Most Intelligent

Fastest

Lowest Price

Intelligence