INDEPENDENT TELEMETRY • Q4 2026

LLM Performance, Speed & Pricing Index

Objective side-by-side comparison across intelligence quality, coding reliability, token throughput (t/s), and blended API cost (3:1 input/output ratio).

Tip: Click any column header to sort by price, speed, or ROI.

Top 5 by Output Speed (tokens / sec)

Gemini 2.5 Flash (Google)215 t/s
Llama 4 Maverick (Meta)185 t/s
GPT-4.1 Mini (OpenAI)168 t/s
OpenAI o3-mini (OpenAI)126 t/s
Gemini 2.5 Pro (Google)112 t/s

Top 5 by ROI Value Score (Quality / Cost)

Gemini 2.5 Flash ($0.26/1M)99 / 100
DeepSeek R1 ($0.96/1M)98 / 100
DeepSeek V3 ($0.48/1M)97 / 100
Qwen3 235B MoE ($0.70/1M)96 / 100
Gemini 2.5 Pro ($3.44/1M)95 / 100
Model & Best UseContextDeploy
#1Gemini 2.5 Flash
Google • High-Volume Agents & Real-Time UI
88.987.5215 t/s
TTFT 0.28s
$0.26
$0.15 / $0.6
1M99Use API
#2DeepSeek R1OPEN
DeepSeek • High-ROI Reasoning & Math Workflows
92.693.264 t/s
TTFT 0.95s
$0.96
$0.55 / $2.19
128K98Use API
#3DeepSeek V3OPEN
DeepSeek • Cost-Effective Bulk Content & Translation
89.490.178 t/s
TTFT 0.61s
$0.48
$0.27 / $1.1
128K97Use API
#4Qwen3 235B MoEOPEN
Alibaba • Bilingual Enterprise & Private Deployment
91.591.894 t/s
TTFT 0.49s
$0.70
$0.4 / $1.6
128K96Use API
#5Gemini 2.5 Pro
Google • Long-Context Analysis & Multimodal RAG
95.295.1112 t/s
TTFT 0.65s
$3.44
$1.25 / $10
1M95Use API
#6Llama 4 MaverickOPEN
Meta • Self-Hosted Inference & Fine-Tuning
88.186.2185 t/s
TTFT 0.31s
$0.38
$0.22 / $0.85
1M94Use API
#7GPT-4.1 Mini
OpenAI • Customer Support & SaaS Classification
87.286.8168 t/s
TTFT 0.34s
$0.70
$0.4 / $1.6
1M93Use API
#8GPT-4.1
OpenAI • Structured Extraction & Tool Calling
93.494.0105 t/s
TTFT 0.54s
$3.50
$2 / $8
1M91Use API
#9OpenAI o3-mini
OpenAI • Algorithmic Reasoning & STEM Audit
91.092.5126 t/s
TTFT 0.88s
$1.93
$1.1 / $4.4
200K90Use API
#10Claude 4 Sonnet
Anthropic • Production Coding & Autonomous Agents
94.896.488 t/s
TTFT 0.72s
$6.00
$3 / $15
200K89Use API