Polylog
← Trends

Frontier Model Price War

Frontier API vendors increasingly compete on strategic price-cutting rather than pure capability, repeatedly launching or repricing models below prevailing rates to grab share and compressing industry inference margins; expect recurring below-rival pricing moves.

weakening · confidence 98 · +4 7d · +6 30d · Short term (next 30 days) · tracking since July 31, 2026 · updated September 14, 2026

Sign in to get threshold and movement alerts for this trend.

Score history

Daily conviction score, 0 to 100. Higher means the thesis is more strongly corroborated.

Sep 13 · 100Sep 14 · 98

Now 98 · -2 since Sep 13 · ranged 98 to 100

Showing the last few days. Unlock full score history.

Why the conviction moved

  • Sep 13
    Strengthened +7

    Meta listed Muse Spark 1.1 at $1.25 per million input tokens against $10 for Claude Fable 5.1, an eight-fold undercut on input pricing at comparable context length. A new entrant anchoring at a fraction of the incumbent rate forces the price floor down independent of capability parity.

  • Sep 12
    Strengthened +4

    Sakana AI split its router into a cheap tier and a capability tier, with Fugu Max at $2 per million input and $6 per million output tokens. Competing on blended routed price rather than a single model's quality gives vendors a way to undercut frontier rates without matching frontier training spend.

  • Sep 12
    Strengthened +5

    Anthropic cut Claude Fable 5.1 cache reads by 75 percent to $0.25 per million tokens, which it says lowers typical workload cost by about a quarter and heavily agentic workloads by up to roughly 45 percent. The cut targets the input side of agent loops rather than headline per-token rates, showing price competition moving to the specific cost lines that decide whether long-running agents are economic.

Showing the last 2 days. Unlock the full record.

Source trail

  • Supporting · September 13, 2026

    Meta Puts Muse Spark 1.1 Behind a Metered API and Undercuts Anthropic on Input Tokens

    Meta listed Muse Spark 1.1 at $1.25 per million input tokens against $10 for Claude Fable 5.1, an eight-fold undercut on input pricing at comparable context length. A new entrant anchoring at a fraction of the incumbent rate forces the price floor down independent of capability parity.

    Meta AI

  • Supporting · September 12, 2026

    Anthropic Cuts Claude Fable 5.1 Cache Reads by 75 Percent

    Anthropic cut Claude Fable 5.1 cache reads by 75 percent to $0.25 per million tokens, which it says lowers typical workload cost by about a quarter and heavily agentic workloads by up to roughly 45 percent. The cut targets the input side of agent loops rather than headline per-token rates, showing price competition moving to the specific cost lines that decide whether long-running agents are economic.

    Anthropic

Unlock full source trail, score history, and daily updates.

40 more sources in the full trail.

Unlock Trends

Affected regions & assets

RegionsUnited States
Assets4 assetsUnlock Trends

Townsquare

Argue the thesis in Townsquare.