LIVE
ANTHROPICOpus 4.7 benchmarks published2m ago
CLAUDEOK142ms
OPUS 4.7$15 / $75per Mtok
CHATGPTOK89ms
HACKERNEWSWhy has not AI improved design quality the way it improved dev speed?14m ago
MMLU-PROleader Opus 4.788.4
GEMINIDEGRADED312ms
MISTRALMistral Medium 3 released6m ago
GPT-4o$5 / $15per Mtok
ARXIVCompositional reasoning in LRMs22m ago
BEDROCKOK178ms
GEMINI 2.5$3.50 / $10.50per Mtok
THE VERGEFrontier Model Forum expansion announced38m ago
SWE-BENCHleader Claude Opus 4.772.1%
MISTRALOK104ms
ANTHROPICOpus 4.7 benchmarks published2m ago
CLAUDEOK142ms
OPUS 4.7$15 / $75per Mtok
CHATGPTOK89ms
HACKERNEWSWhy has not AI improved design quality the way it improved dev speed?14m ago
MMLU-PROleader Opus 4.788.4
GEMINIDEGRADED312ms
MISTRALMistral Medium 3 released6m ago
GPT-4o$5 / $15per Mtok
ARXIVCompositional reasoning in LRMs22m ago
BEDROCKOK178ms
GEMINI 2.5$3.50 / $10.50per Mtok
THE VERGEFrontier Model Forum expansion announced38m ago
SWE-BENCHleader Claude Opus 4.772.1%
MISTRALOK104ms

Datasets

Public daily-snapshot mirrors of the TensorFeed API on Hugging Face. Each snapshot is a point-in-time JSONL artifact suitable for RAG, evaluation, agent context, and time-series analysis. Inference-only license consistent with the AFTA standard.

tensorfeed / ai-ecosystem-daily

Daily JSONL snapshots of the entire public TensorFeed API, committed at 08:00 UTC via GitHub Actions. 36 feeds per day, ~900 records per snapshot.

Cadence
Daily 08:00 UTC
Format
JSONL per feed
License
Inference-only
Cost
Free

Quick start

from datasets import load_dataset

# Load any single feed
news = load_dataset("tensorfeed/ai-ecosystem-daily", "news", split="train")
models = load_dataset("tensorfeed/ai-ecosystem-daily", "models", split="train")
gpu = load_dataset("tensorfeed/ai-ecosystem-daily", "gpu-pricing", split="train")

# Filter by date (filename pattern is YYYY-MM-DD/)
recent = news.filter(lambda x: x["fetchedAt"] >= "2026-05-01")

36 feeds per daily snapshot

Each feed is a JSONL file under YYYY-MM-DD/. Configs are loadable individually via load_dataset(repo, "feedname").

news
up to 200 rec
AI news articles, snippets clipped, prompt-injection sanitized
models
~230 rec
Model pricing and specs, flattened with provider on each row
pricing
1 rec
Compact pricing payload for agents
status
~12 rec
Real-time operational status of major AI services
benchmarks
~5 rec
Public benchmark scores per model
agents-directory
~18 rec
Curated AI agent directory
agents-activity
varies rec
Live AI bot traffic on TensorFeed
podcasts
~50 rec
Recent AI podcast episodes
trending-repos
~20 rec
Trending GitHub repos in AI/ML
mcp-registry
1 rec
Daily count and 1-day delta of the official MCP registry
probe
1 rec
Last 24h of LLM endpoint latency probes
gpu-pricing
1 rec
GPU rental price snapshot across cloud marketplaces
afta-adopters
varies rec
Sites publishing an AFTA manifest
ai-hardware
~17 rec
AI accelerator specs (NVIDIA, AMD, TPU, Trainium, etc)
open-weights
~9 rec
Open-weights models with quantization + VRAM requirements
inference-providers
~8 rec
Hosted inference for open-weights
training-runs
~11 rec
Disclosed and estimated training cost catalog
marketplaces
~12 rec
AI marketplace catalog
specialized-models
~19 rec
Domain-specialized models
fine-tuning
~12 rec
Fine-tuning providers (first-party and hosted)
oss-tools
~25 rec
Production OSS tools agents install
agent-apis
~29 rec
Non-LLM APIs in the agent stack
voice-leaderboards
1 rec
TTS Arena Elo + Open ASR Leaderboard WER
embeddings
~18 rec
Embedding and reranker model catalog
multimodal
~24 rec
Image, video, TTS, STT model catalog
vector-dbs
~12 rec
Vector database catalog
frameworks
~15 rec
Agent framework catalog
benchmark-registry
~24 rec
Benchmark catalog with active/saturated status
public-leaderboards
~20 rec
Pointers to every live public AI leaderboard
conferences
~18 rec
AI conferences with dates
funding
~21 rec
AI funding rounds catalog
model-cards
~8 rec
Per-model system / safety / red-team docs
ai-policy
~10 rec
AI regulation tracker
compute-providers
~17 rec
GPU cloud, hyperscaler, AI-serverless catalog
usage-rankings
~20 rec
Model usage rankings
agent-provisioning
~18 rec
Agent infrastructure providers

Inference-only license

The dataset is released under TensorFeed's inference-only license. You may use it as input context for AI agents and LLM inference: RAG, evaluation, prompt context, agent toolchains. You may not use it as training data for foundation models without explicit written permission.

The license is part of the Agent Fair-Trade Agreement: the same standard that governs paid API access on tensorfeed.ai. Compliant agents get a perpetual usage right; non-compliant training pipelines do not.

If daily is too slow

The dataset is a daily mirror. The live API is updated continuously: news every 10 minutes, status every 5 minutes, models and benchmarks daily, GPU pricing every 4 hours.