Nightly automated scraping

Every LLM price change.
Captured and timestamped.

TokenScope scrapes every major provider's pricing pages nightly, detects changes the moment they happen, and maintains a complete public changelog with price history — so you're never caught off guard by a rate card update buried in the docs.

Live Price + MMLU benchmark
DeepSeek V4 Lite $0.1400/M MMLU 86.10
o4-mini $0.5500/M MMLU 93.70
Gemini 3.0 Flash $0.1000/M MMLU 87.20
Claude Sonnet 4 $3.0000/M MMLU 88.10
DeepSeek R1 $0.5500/M MMLU 96.40
Updated 2026-05-31 03:00 UTC View benchmarks →

How it works

Automated. Comprehensive. Real.

01

Scrape every night

Every provider's official pricing and docs pages fetched at 03:00 UTC. No manual updates, no stale data.

02

Detect changes instantly

Diff engine compares last run against current. Any rate card change gets flagged and timestamped in the changelog.

03

Publish to public log

Every change, ever. A timestamped record of the price war as it happened — input, output, cache-hit rates for all models.

04

Generate comparison pages

Auto-built pages for every model and every head-to-head. GPT-5 vs Claude. Cheapest model for code. You name it.

Providers tracked

Every major LLM API, covered.

OpenAI
GPT-5.5, o3, o4-mini
from $0.05/M in
Anthropic
Claude Opus 4.6, Sonnet 4.6
from $1.00/M in
Google
Gemini 3.1 Pro, 3 Flash
from $0.30/M in
DeepSeek
V4 Flash, V4 Pro, R1
from $0.14/M in
Groq
Llama 3.3, Qwen, Mistral
from $0.05/M in
Mistral
Large 3, Small 3.2
from $0.10/M in
xAI
Grok 4.1, 4.1 Fast
from $0.20/M in
Together AI
Llama 4, DeepSeek V4
from $0.05/M in
Cohere
Command R7B, R+ 08-2024
from $0.04/M in
Fireworks
Llama 4, Qwen, Mixtral
from $0.05/M in
Perplexity
Sonar Pro, Sonar
from $0.15/M in
OpenRouter
500+ models, unified
from $0.00/M in

Recent changes

The public changelog

2026-05-30
DeepSeek V4 Flash input ▼ $0.14 → $0.14
Cache-hit rate dropped 90%: $0.028 → $0.0028/M tokens. Massive for agent workloads.
2026-05-28
Anthropic Claude Opus 4.6 input ▼ $15 → $5/M
67% price cut from Opus 4.1 pricing. Now competitive with Gemini 3.1 Pro.
2026-05-26
Google Gemini 3.1 Flash-Lite NEW
$0.25/M input, $1.00/M output. Competes directly with DeepSeek V4 Flash on price.
2026-05-24
OpenAI GPT-5.4 mini ▼ $0.40 → $0.20/M input
50% cut. Now cheaper than GPT-4.1 nano on input, same context window.
2026-05-20
xAI Grok 4.3 NEW
$0.50/M in, $2.00/M out. New flagship from xAI, competitive with Gemini 3 Flash.
Full history, price charts, and all model pages — always up to date.
Built by Polsia

The LLM price war never stops.
Now you can watch it live.

TokenScope tracks every move so you don't have to. Accurate pricing data, updated automatically, forever free to access.