6 replies

Telegram streaming not working with Ollama — messages sent all at once

Version: OpenClaw 2026.2.15 | OS: Linux | GPU: RTX 4080
Model: gptoss20b-65k (20B) via Ollama | Channel: Telegram (group w/ forum topics)

Problem

streamMode: "partial"

streamMode: "partial"

not working. Messages arrive all at once after generation instead of being progressively edited.

Tested in DM, group with forum topics, and WebChat — same result everywhere.

What I verified

- Ollama streams correctly via curl (

stream: true

stream: true

, token-by-token ~140ms)
- Tried both

"api": "ollama"

"api": "ollama"

and

"api": "openai-completions"

"api": "openai-completions"

— same result
-

/reasoning stream

/reasoning stream

enabled
- Long responses (500+ words) also arrive all at once
- Logs show run starts and Ollama generates, but no progressive edits happen

Logs

agent/embedded: run start provider=ollama_cpu model=gptoss20b-65k:latest thinking=off messageChannel=telegram
diagnostic: run registered totalActive=1

agent/embedded: run start provider=ollama_cpu model=gptoss20b-65k:latest thinking=off messageChannel=telegram
diagnostic: run registered totalActive=1

Ollama confirms streaming works on its side. No Telegram edits occur — full message sent only after completion.

Config (I have some settings here just to test)

"telegram": {
  "enabled": true,
  "dmPolicy": "open",
  "groups": { "*": { "requireMention": false } },
  "groupPolicy": "open",
  "streamMode": "partial"
}

"telegram": {
  "enabled": true,
  "dmPolicy": "open",
  "groups": { "*": { "requireMention": false } },
  "groupPolicy": "open",
  "streamMode": "partial"
}

"ollama_cpu": {
  "baseUrl": "http://127.0.0.1:11434",
  "api": "ollama",
  "models": [{
    "id": "gptoss20b-65k:latest",
    "reasoning": false,
    "contextWindow": 65536
  }]
}

"ollama_cpu": {
  "baseUrl": "http://127.0.0.1:11434",
  "api": "ollama",
  "models": [{
    "id": "gptoss20b-65k:latest",
    "reasoning": false,
    "contextWindow": 65536
  }]
}

Anything else needed for streaming to work with Ollama on Telegram? Or is this a known limitation?

FotC�

Friends of the Crustacean 🦞🤝•4w ago•

6 replies

dropa200

Telegram streaming not working with Ollama — messages sent all at once

Version: OpenClaw 2026.2.15 | OS: Linux | GPU: RTX 4080
Model: gptoss20b-65k (20B) via Ollama | Channel: Telegram (group w/ forum topics)

Problem

streamMode: "partial"

streamMode: "partial"

not working. Messages arrive all at once after generation instead of being progressively edited.

Tested in DM, group with forum topics, and WebChat — same result everywhere.

What I verified

- Ollama streams correctly via curl (

stream: true

stream: true

, token-by-token ~140ms)
- Tried both

"api": "ollama"

"api": "ollama"

and

"api": "openai-completions"

"api": "openai-completions"

— same result
-

/reasoning stream

/reasoning stream

enabled
- Long responses (500+ words) also arrive all at once
- Logs show run starts and Ollama generates, but no progressive edits happen

Logs

agent/embedded: run start provider=ollama_cpu model=gptoss20b-65k:latest thinking=off messageChannel=telegram
diagnostic: run registered totalActive=1

agent/embedded: run start provider=ollama_cpu model=gptoss20b-65k:latest thinking=off messageChannel=telegram
diagnostic: run registered totalActive=1

Ollama confirms streaming works on its side. No Telegram edits occur — full message sent only after completion.

Config (I have some settings here just to test)

"telegram": {
  "enabled": true,
  "dmPolicy": "open",
  "groups": { "*": { "requireMention": false } },
  "groupPolicy": "open",
  "streamMode": "partial"
}

"telegram": {
  "enabled": true,
  "dmPolicy": "open",
  "groups": { "*": { "requireMention": false } },
  "groupPolicy": "open",
  "streamMode": "partial"
}

"ollama_cpu": {
  "baseUrl": "http://127.0.0.1:11434",
  "api": "ollama",
  "models": [{
    "id": "gptoss20b-65k:latest",
    "reasoning": false,
    "contextWindow": 65536
  }]
}

"ollama_cpu": {
  "baseUrl": "http://127.0.0.1:11434",
  "api": "ollama",
  "models": [{
    "id": "gptoss20b-65k:latest",
    "reasoning": false,
    "contextWindow": 65536
  }]
}

Anything else needed for streaming to work with Ollama on Telegram? Or is this a known limitation?

Telegram streaming not working with Ollama — messages sent all at once

Problem

What I verified

Logs

Config (I have some settings here just to test)

Similar Threads

Telegram streaming not working with Ollama — messages sent all at once

Problem

What I verified

Logs

Config (I have some settings here just to test)

Similar Threads

Similar Threads

Similar Threads