Eight models. One endpoint.

• Synapse Embed

  syn-embedembeddings

  1,536-dimension embeddings for search and retrieval.

  context

  8K

  in / 1M

  $0.02

  out / 1M

  —

  speed

  9,000 tok/s

  first token

  40ms

  fine-tune

  no

  Use syn-embedEstimate cost

• Synapse Mini

  syn-minitext

  Classification, routing and extraction at a fraction of a cent.

  context

  32K

  in / 1M

  $0.04

  out / 1M

  $0.16

  speed

  640 tok/s

  first token

  90ms

  fine-tune

  yes

  reasoning eval 61

  code eval 48

  Use syn-miniEstimate cost

• Synapse Flash

  syn-flashtextvision

  Our fastest general model. The default for chat and agents.

  context

  128K

  in / 1M

  $0.10

  out / 1M

  $0.40

  speed

  420 tok/s

  first token

  120ms

  fine-tune

  yes

  reasoning eval 74

  code eval 69

  Use syn-flashEstimate cost

• Synapse Vision Lite

  syn-vision-litevision

  Cheap OCR, captioning and document parsing.

  context

  64K

  in / 1M

  $0.25

  out / 1M

  $1.00

  speed

  300 tok/s

  first token

  160ms

  fine-tune

  no

  reasoning eval 64

  code eval 41

  Use syn-vision-liteEstimate cost

• Synapse Code

  syn-codetextcode

  Tuned for completion, refactors and repository-scale context.

  context

  256K

  in / 1M

  $0.60

  out / 1M

  $2.40

  speed

  260 tok/s

  first token

  150ms

  fine-tune

  yes

  reasoning eval 78

  code eval 88

  Use syn-codeEstimate cost

• Synapse Pro

  syn-protextvision

  Balanced reasoning and vision for production workloads.

  context

  200K

  in / 1M

  $1.20

  out / 1M

  $4.80

  speed

  160 tok/s

  first token

  210ms

  fine-tune

  yes

  reasoning eval 86

  code eval 82

  Use syn-proEstimate cost

• Synapse Voice

  syn-voiceaudio

  Speech in, text out. Streaming transcription with speaker turns.

  context

  32K

  in / 1M

  $2.00

  out / 1M

  $4.00

  speed

  220 tok/s

  first token

  140ms

  fine-tune

  no

  reasoning eval 58

  code eval 30

  Use syn-voiceEstimate cost

• Synapse Ultra

  syn-ultratextvisioncode

  Our most capable model, with a one-million-token window.

  context

  1M

  in / 1M

  $4.00

  out / 1M

  $16.00

  speed

  75 tok/s

  first token

  380ms

  fine-tune

  no

  reasoning eval 93

  code eval 90

  Use syn-ultraEstimate cost

Blended price assumes three input tokens per output token. Evals are Synapse internal benchmarks (0 to 100), run on the same harness for every model.