Network

Live pricing across the AntSeed P2P inference network — every provider, every advertised model, ranked by rate.Updated 22:23

49 active peers · 48 on-chain proven · from our own DHT crawl
Services
864
advertised (peer × model)
Median saving
−56%
vs list price · 297 at half price or less · cost calculator →
Free services
97
$0 in & out · filter →
Active peers
49
48 on-chain proven
Tokens served
138.89B
network-wide, on-chain
Models
239
distinct, advertised
Service directory
Filters
Labs
Providers
4 services (of 864) · most-used models first · Tokens & Users are provider-level totals · marks services whose buyers share a per-model split on-chain (opt-in, partial) — shown under the totals
Service Input $/MCached $/MOutput $/MSaveReputationVerifiedStakeTokensUsers
Qwen 3.5 397B
Qwen 3.5 397B
▲ Apex AntChatCodingLong context
$0.34$0.34$2.0338%92.14$101.0011.57B124
Qwen 3.5 397B
Qwen 3.5 397B
Venice.ai ProxyChatCodingReasoning
$0.375$0.375$2.2532%100.00$10.005.83B
1.64k
159
1
$0.4292$0.4292$2.575122%41.29$10.00243.37M54
Qwen 3.5 397B
Qwen 3.5 397B
PhalaChatReasoningVision
$0.55$0.225$3.5044.17$10.00626.77M
30.16M
13
3
ChatCodingLong context
Input $/M$0.34
Cached $/M$0.34
Output $/M$2.03
Save38%
Reputation92.14
Verified
Stake$101.00
Tokens11.57B
Users124
ChatCodingReasoningVision
Input $/M$0.375
Cached $/M$0.375
Output $/M$2.25
Save32%
Reputation100.00
Verified
Stake$10.00
Tokens5.83B
1.64k
Users159
1
Input $/M$0.4292
Cached $/M$0.4292
Output $/M$2.5751
Save22%
Reputation41.29
Verified
Stake$10.00
Tokens243.37M
Users54
ChatReasoningVision
Input $/M$0.55
Cached $/M$0.225
Output $/M$3.50
Save
Reputation44.17
Verified
Stake$10.00
Tokens626.77M
30.16M
Users13
3

Live pricing/services directory from our own AntSeed DHT crawl. Prices are advertised $/1M tokens, not settled amounts. "Reputation" = on-chain trust (0-100), "Verified" = signed DHT announcement, "Stake" = seller USDC stake on-chain. "Tokens" and "Users" are PROVIDER-level on-chain totals shown on every one of that provider's model rows: the per-purchase model is not recorded on-chain, so usage cannot be split per model. Peers marked "new" advertise on the DHT but have no on-chain settlement yet.